Measured

Prof. L MakhubuPresidentTWOWSTrieste -ItalyPhone: (+39) 040 224 0321Fax; (+39) 040 224 0689,Radiation Protection InstituteGhana Atomic Energy CommissionP. O. Box LG 80LegonAccraPhone: (+23321) 400 310Fax: (+23321) 400 807E-mail: lI13S3SSiamah@y.co.uk20 th May 2005Dear Prof. Makhubu,POSTAGE OF THESISPlease find attached a copy of my PhD thesis to lWOWS for your records. I would like to take theopportunity to express my appreciation for all the support received.Thanks for your kind consideration.Please acknowledge receipt.Yours faithfullyMary Assiamah (Ms.)

TWWS - Acknowledgement ofreceipt"Subject: TWOWS - Acknowledgement ofreceiptFrom: TWOWS Date: Mon, 27 Jun 2005 13:30:46 +0200To: massiamah@yahoo.co.ukDr. Mary AssiamahGhana Atomic Energy CommissionP.o. Box 80Legan, AccraGhanaDear Dr. Assiamah,We wish to acknowledge receipt of the Ph.D. thesis on • Dosimetric Techniques forMammography Mass Screenin~ Programs".Congratulations on the successful completion of your Ph.D. research program.Wishing you all the best for your future endeavours.With best regards,TWOWS SecretariatThird World Organization for Women in Science (TWOWS)c/o TWASThe Abdus Salam International Centre for Theoretical Physicspostal address: Strada Costiera 11 - 34014 Trieste - ItalyFax: (+39 040 2240-689)E-mail: info@twows.orghttp://W\~_twows.org(ICTP)1of!612712005 1:30 PM

DOSIMETRIC TECHNIQUES FOR MAMMOGRAPHYMASS SCREENING PROGRAMSMary Assiamaho9Submitted to the Faculty of Science, University ofthe Witwatersrand,Johannesburg, South Africa, in fulfilment ofthe requirement for the degree ofDoctor ofPhilosophyNovember 2004

DECLARATIONI declare that this thesis is my own, unaided work and it has not beensubmitted before for any degree or examination in any other University.It is being submitted for the degree of Doctor of Philosophy in theUniversity ofthe Witwatersrand, Johannesburg, South Africa.-----day of--------,2004II

ABSTRACTScreening of asymptomatic women using X-ray mammography technique is very commonin many parts of the world in view of the prevalence of breast cancer among women.Mammography X-ray procedures are well established; with radiation dose measurementsusually carried out using air ionisation chamber despite its inherent disadvantages. In thisstudy, the various parameters necessary for accurate dose calculations from manunographyX-ray energies and their effect on the calculated dose values, the relationship between dose,breast size, image quality and X-ray tube parameters as well as an alternative method fordose measurements, were systematically investigated.A method is presented for calculating accurately the mass attenuation and mass-energycoefficients for any energy bin of interest in the photon energy region 1-20 keV fromexisting mass attenuation and mass-energy coefficients data. Data fitting procedure wasused for the study using an established equation. The results of the study showed that whendata points containing high and low energies such as I - 200 keY are fit together with asingle set of parameters, an overestimation of about 20% at the lower energies with fargreater deviations at higher energies can result. It has been shown that grouping data intosmaller energy regions when fitting would lead to accurate calculations of the massattenuation or mass energy-absorption data. This is especially important if the data were tobe used in low energy photon calculations such as would be the case for manunographybeams.III

An investigation into the effect of pressure, temperature and humidity in air on photonfluence at a typical mammography, low bremsstrahlung energy (25 kVp), has been carriedout. The results of the investigation showed that air kerma values from an X-ray spectrumthat has significant lower energy components is likely to be more sensitive to changes inpressure, temperature and humidity than the air kerma from an X-ray spectrum with lowerenergy components less pronounced.Mean glandular dose (MGD) values had been calculated for various tube potentials andtube loadings (TL) using direct measurements of the incident entrance air kerma (ESAK) atthe surface of a standard breast phantom and also from spectral measurements acquiredwith a solid-state detector. Detailed presentations of dose measurements from directmeasurements and also from X-ray spectral data employing the established methods aregiven. Comparisons of the MGD values thus derived are presented and the relationshipbetween MGD, phantom thickness,image quality and tube operating parameters isdiscussed.The possibility of evaluating radiation dose from mammography X-ray beams usingconstructed probes with diamond as the active radiation sensing material has been studied.Diamond has been used in the conduction mode whereby electrodes are connected to it andthe resulting current from the interaction of the ionizing radiation with the diamonddetected. Single crystal diamonds produced under high pressure and temperature (HPHT),as well as polycrystalline diamonds manufactured by the chemical vapour deposition(CVO) method were used. Suitable diamond stones were carefully selected for the studyusing various techniques. The probe was constructed entirely using tissue equivalentIV

materials. In current practice diamond in the form of thin plates are used in the "flat-on"geometry, where the radiation beam to be monitored or measured, impinges on the flat faceof the diamond. In this work it was found that using diamond plates in a side-on, or "edgeon",geometry improves the collection efficiency ofthe diamond.The probe has been designed for radiation detection in both "edge-on" and "flat-on" sensorgeometry profiles without having firstly to unseat the diamond sensor element from itsoriginal position within the probe housing before taking measurements. The study hasshown that with the "edge-on" geometry configuration, radiation from impinging photonswith energies below 30 keV can be made to deposit almost all (about 90%) of their energyinto the sensor. The probe was designed for use in combination with commerciallyavailable electrometer systems. The response of the diamond probe to changes in radiationdose correlated well with that obtained from a secondary standard ionization chamber at thesame X-ray tube settings.v

1 dedi¢te t4i4 tke4i4 t "'if ~IMe«a~a404e~,

ACKNOWLEDGEMENTSI would like to render my deepest gratitude to the One lowe allegiance to, my Shield, myStrength, my Shelter, my Portion, Strong Tower, and my very Present Help in times ofneed, the Almighty God, the Great Jehovah and Provider, without whom this thesis wouldnot have seen the light ofday, to Him be all the glory and honour throughout eternity.My profound gratitude goes to my supervisors, Professor Thomas Leong Nam andProfessor Emeritus Rex James Keddy for their guidance and encouragement. Their extremeavailability and continuing support towards this work will always be appreciated.To the following staff members of Schonland Research Institute for Nuclear Sciences,University ofthe Witwatersrand, I would also like to acknowledge my appreciation:• Professor Trevor Derry for the role he played in my admission into Wits; ProfessorJohn \Vatterson, Professor John Carter, Professor BaIt Verhagen, Doctor SimonConnell, Doctor Roger Hart and Doctor Ellias Sideras Haddad for their kindnessand encouragement.• Mr Trevor Hollander and Mr Richard Chilwa for their valuable inputs into theconstruction of the diamond probes as well as their general advice and assistance;Mr Eric Rood and Mr Oleg Parker for all the technical assistance they provided; MrMarias Labuschagne, for constructing the diamond probes; Mr Mick Rebak for hisskilful polishing of the diamond samples and helpful suggestions; Mr Johan Geyerfor his kind assistance and continual support.• Mrs Doris Monyamane whose cheerful demeanour and her tirnely encouragementssent through her regular e-mail messages made an indirect nonetheless positiveimpact on this work. Her support will always be appreciated. Thanks also to MrsMarina Labuschagne and Mrs Cynthia Hlatshwayo for all the assistance provided.Vll

My profound thanks also go to the following:Dr R.J. Caveney of Wits Enterprise far his support and help.Dr Andrew Whitehead ofElement Six Ltd for the supply ofthe CVD diamondspecimens used in this study.Dr Robbie Burns ofElement Six (Pty) Ltd for providing the HPHT highnitrogen crystals.Dr Graeme Hill of De Beers Group Services (Pty) Ltd for the use ofthe low energyhigh purity germanium detector system for X-ray spectral measurements.I would also like to express my sincere thanks and appreciation to the Third WorldOrganization for Women in Science (TWOWS) for sponsoring me throughout the durationof this study. The generous financial support received enabled me to focus and concentrateon my work.To my employer, the Ghana Atomic Energy Commission, I would like to say thank youvery much for granting me the permission to spend these years at the University of theWitwatersrand in pursuit of my postgraduate studies.Lastly, but by no means the least I would like to thank my family for their unflinchingsupport and encouragement.IX

CONTENTTitle page .Declaration........................................................ 11Abstract... ... .. ... ..... . .. .. .... .. ... .. .. .. ... .. .. 111Dedication.............................................................................................Acknowledgements...............Table ofcontents............................................VIVIIxCHAPTER ONE. .. .. .. . .. ... . .. ... .. .. ... .. ..... .... . .. .. I - 27General IntroductionCHAPTER TWO .28 - 54Comparison of mammography radiation dose values obtained from direct incident airkerma measurements with values from measured X-ray spectral dataAcceptedfor publication in Applied Radiation and IsotopesDosimetric techniques for mammography X-ray beams............................. ..... 55 - 56Radiation Physics and Chemistry 71 (2004) 957-958CHAPTERTHREE 57-63Segmented multifit of polynomial function for mass attenuation and energy-absorptioncoefficient valuesRadiation Physics and Chemistry 67 (2003) 1-6CHAPTER FOUR.......................................................................... 64 - 78Effect of pressure, temperature and humidity in air on photon f1uence and air kerma valuesat low photon energiesRadiation Physics and Chemistry 68 (!003) 707-720CHAPTERFIVE 79-114Synthetic diamond as a radiation sensing element for mammography X-ray beam dosemeasurementsx

CHAPTERSIX 115 -151Comparative studies of measured mammography X-ray beam doses with theoreticalcalculations from the Monte Carlo code system PENELOPE using synthetic diamond asradiation sensorCHAPTER SEVEN 152 - 157General ConclusionsAPPENDICES.............................................................................. 158 - 165Xl

Chapter 1GENERAL INTRODUCTION

1.1. Historical BackgroundMammography is used in two main categories: screening and diagnosis. Screening involvesexamination ofasymptomatic women with the aim ofdetecting breast lesions at an early stagebefore the lesion becomes palpable. Diagnostic mammography on the other hand is done onwomen who through physical findings or the symptoms, may show that they are considered toalready have breast cancer. The use ofX-ray mammography in the screening ofasymptomaticwomen has become very common in many parts of the world, in view of the fact that breastcancer has been found to be one of the most common malignant diseases of women. Breastcancer is reported to be the second highest cause of cancer deaths in women today (after lungcancer), with an estimation of 400,600 deaths from breast cancer for 2001 (GLOBOCAN,2000). In 2001, World Health Organization (WHO) predicted more than 1.2 million newbreast cancer cases (globally) while in United States, both the American Cancer Society andthe National Cancer Institute estimated approximately 192,200 new cases of invasive breastcancer cases for 2001. Male breast cancer has also been reported but this accounts for less than1% of the total (American Cancer Society, 2001; National Cancer Institute, 1999). Thereported death rates from breast cancer declined significantly between 1992 and 1996.Statistics on mammography indicate that the incidence of breast cancers per 100,000 womenincreased by approximately 4% during the 1980s and levelled off to about 100 cases per100,000 women in the 1990s. Currently over 900,000 new breast cancer cases are registeredannually worldwide (Siemens, 2004). Early diagnosis of breast cancer plays a leading role inimproving the patient's prognosis. This is accomplished globally by mass screening ofhealthywomen using mammography X-ray techniques. The objective of the screening exercise is todetect breast cancer at an early stage to reduce breast cancer mortality. Clinical studies haveproven that with early detection and proper patient management the mortality rate is reduced2

(Siemens, 2004). In the screening programs, it is essential that low radiation dose protocols beused to reduce cancer induction by radiation to an absolute minimum, and achieve qualityhealth: the ultimate aim ofmedical services.1.2. Mammography TechniqueMammography is the most sensitive technique currently available for the early detection ofbreast lesions and therefore is the method of choice (Fahrig and Yaffe, 1994a; Simonetti et al1998). Mammography is a radiographic technique that uses X-rays produced from amolybdenum, rhodium or tungsten or combination of any two of them as anode or targetmaterials with an exit window of beryllium or glass. Some older machines have glass as exitwindow. The mammography X-ray beam is generated when high-velocity electrons from thecathode collide with the target or anode material. Only one percent of the energy of thestreams of electrons is transformed directly into X-ray energy producing bremsstrahlung andcharacteristic X-rays. The remaining ninety-nine percent of the energy of the electrons isconverted into heat and dissipated away. To avoid excessive heating ofone point of the anodematerial surface and cause irreparable damage to the anode material, most modernmammography machines have rotating anodes. The X-ray tube is mounted on tube housingwhich provides mechanical support and also serves as a container to store oil used to cool theanode during operation. The mammography machine is designed such that the generated X-ray beam is directed through a window by a primary beam restrictor and then passed through afilter l . The filter materials commonly used in mammography are molybdenum, rhodium andaluminium. The filtered beam is collimated to desired dimensions for patient imaging.1 A material insertedin the X-ray beam to alterthe qualityofthe beam by removing unwanted energy which doesnot contribute to image information but increases patientdose.3

Xeromammography and screen-film mammography are the two main mammography imagingtechniques most commonly used for the detection and diagnosis ofbreast cancer.1.2.1. XeromammographyMost xeromammography units employ ceiling-mounted X-ray tubes with tungsten andaluminium as anode or target and filter materials respectively. Xeromammography techniquesare usually operated at peak kilovoltage (kVp) ranging between 40 and 55 kVp and a tubecurrent of 300 milliampere (rnA). The image receptor is a thin sheet of photoconductingamorphous' selenium. Dedicated Or specialised mammography X-ray machines can also beused in xeromammography with aluminium as filter material instead of molybdenum filtrationwhich is employed in screen-film mammography. In xeromammography, the X-ray images ofthe breast are recorded on a uniformly charged selenium plate held in a light-proof cassette.Upon X-ray exposure, the incident X-ray beam selectively dissipates the charge on theselenium plate to form a latent image which is made visible by dusting finely dividedthermoplastic powder onto the plate. To form a permanent copy of the image, the visibleImage IS electrostatically transferred to a plastic-coated paper and followed with thermalbonding.1.2.2 Screen-film mammographyScreen-film mammography employs a specialised or dedicated mammography X-ray unitwhich consists ofa molybdenum target and molybdenum filter usually of thickness 30 11m. Inscreen-film mammography, the X-ray image of the breast is recorded on a film cassette. TheI An amorphous material is vitreous, glass-like, withthe atoms in random positions.4

Xeromammography and screen-film mammography are the two main mammography imagingtechniques most commonly used for the detection and diagnosis ofbreast cancer.1.2.1. XeromammographyMost xeromammography units employ ceiling-mounted X-ray tubes with tungsten andaluminium as anode or target and filter materials respectively. Xeromammography techniquesare usually operated at peak kilovoltage (kVp) ranging between 40 and 55 kVp and a tubecurrent of 300 milliampere (rnA). The image receptor is a thin sheet of photoconductingamorphous' selenium. Dedicated or specialised mammography X-ray machines can also beused in xeromammography with aluminium as filter material instead of molybdenum filtrationwhich is employed in screen-film mammography. In xeromammography, the X-ray images ofthe breast are recorded on a uniformly charged selenium plate held in a light-proof cassette.Upon X-ray exposure, the incident X-ray beam selectively dissipates the charge on theselenium plate to form a latent image which is made visible by dusting finely dividedthermoplastic powder onto the plate. To form a permanent copy of the image, the visibleimage is electrostatically transferred to a plastic-coated paper and followed with thermalbonding.1.2.2 Screen-film mammographyScreen-film mammography employs a specialised or dedicated mammography X-ray unitwhich consists ofa molybdenum target and molybdenum filter usually of thickness 30 urn. Inscreen-film manunography, the X-ray image of the breast is recorded on a film cassette. TheI An amorphous material is vitreous, glass-like, withtheatoms in random positions.4

mammography cassette consists of a single fluorescent' high-detail intensifying screen inclose proximity with a single-emulsion/ film. The formation of the X-ray image on the filminvolves two steps: (i) the X-ray beam emerging from the patient is transformed into a patternof visible light by the cassette intensifying screen;(ii) the visible image is capturedpermanently on the film. Upon chemical development of the film with a film processor, theemulsion (silver iodobromide crystals) that contains the latent X-ray image centres isconverted into specks ofmetallic silver.For many years, direct exposure with medical X-ray film or industrial film was used formammography until 1972 when the first screen-film combination was designed for thetechnique. Films used in screen-film combinations have higher film contrast and requiresignificantly lower radiation exposure than the direct exposure films (approximately 50 to 100times less) (Barnes and Frey, 1991; Rothenberg and Haus, 1995). Cassettes and screen-filmcombinations used in conventional mammography are designed such that the performance ofthe technique is enhanced. The major components of dedicated mammography X-raymachines are: appropriate beam quality, breast compression device, phototiming or automaticexposure control, focal spot-to film distance, grids and magnification.1.2.2.1. Appropriate beam qualityThe mammography X-ray tube is usually operated at kilovoltage of25 to 32 kVp. Witha molybdenum target, the mammography X-ray beam consists of bremsstrahlung and asignificant component of molybdenum characteristic X-ray photons. The molybdenumI A fluorescent substance emits a pinpoint ofvisible light whenever an incident x-ray photon undergoes aphotoelectric or Compron collision with one of its atoms.2.Emulsion is a suspension of microscopic silver iodobromide crystals in a gelatin.5

target produces characteristic X-ray peaks that occur between 17.5 and 19.6 keY. Thoughnot too penetrating, the characteristic lines provide high subject contrast in tissue resultingin enhanced detection ofcalcifications and image information. To keep exposure times to aminimum, mammographic X-ray tube currents are usually high, at least 100 rnA on a largefocal spot and 80 rnA on a small focal spot. Both rnA values are accurate to ± 10% (Lee eta!., 1995).1.2.2.2. Breast compression deviceThe breast compression device usually a stiff polymethyl methacrylate positioned parallelto the film surface.Good breast compression is an important factor in screen-filmmammography. A number of benefits are associated with good breast compression(NCRP, 1986; Barnes and Frey, 1991) namely:I. Reduction of scattered radiation which reduces significantly the subject contrast.This in turn increases subject contrast leading to improved detection ofcalcification.II.Immobilization of the breast reduces blurring caused by patient motion.111. Reduction in radiation dose.IV.Reduction of geometric unsharpness by locating breast tissue closer to the imagereceptor i.e. reduction ofobject-film distance.v. Production ofmore uniformly thick breast which leads to even the penetration ofthe X-ray beam and a lower difference in radiographic density of the breast areaunder examination.VI.Spreading the breast tissue so that suspicious lesions can be more easily detected.6

1.2.2.3. PhototimingPhototiming, also known as automatic exposure control (AEC), is designed toautomatically provide the radiation exposure needed to produce a mammogram with anacceptable and consistent optical density. Most dedicated mammographic X-ray units haveautomatic exposure control systems. The radiation exposure is normally controlled by aradiation detector located after the image receptor. Exposure is terminated when theradiation dose received by the detector reaches a pre-determined level which correspondsto the desired optical density. For many mammographic units, the position of the radiationdetector can be varied between two or more predetermined positions to facilitate theexposure of breast of differing size and density. To prevent gross over exposure of thebreast, in case ofautomatic exposure control system failure, a guard timer is fitted.1.2.2.4. Focal spot-to-film distanceThe distance between the focal spot' and the X-ray film plays an important role in thesharpness of the image formed. One of the first molybdenum target X-ray units had anominal focal spot diameter of 0.6 mm and focal spot-to-film distance of only 35 em. Ithas been observed that mammography units with larger focal spot sizes and short focalspot-to-film distances have excessive geometric unsharpness". Some mammographic unitshave been designed to produce two focal spots usually 0.3 mm for normal mammographyand 0.1 mm for magnification mammography. To minimize geometric unsharpness orblurring, the focal spot size and object-to-image receptor distance should be minimized,l Focal spot is the region on the x-ray tube anode material's surface from where the X-ray beams are generated.1 Geometric blurring is the lateral spreading ofthe image ofa structural boundary; that is the distance overwhich the optical density changes berween the structure ofinterest and its surrounding takes place (Barnes andFrey, (991).7

while maximizing the focal spot-to-object distance. Modem mammography X-ray unitshave been designed to operate at focal spot-to-film distances of50 cm or more.1.2.2.5. GridTo further reduce scattered radiation and Improve image contrast, dedicatedmammographic X-ray units include grids. Grids are capable ofabsorbing about 50 percentofthe X-rays exiting from the breast. To offset the reduction in X-rays caused by the grids,an increase in the mAs to about 2 to 2.5 times the value for non-grid techniques isrequired. The increase in mAs increases patient radiation dose. Grids are of two types:stationary and moving grids. Stationary grids are ultra-high-strip-density grids havingextremely fine grid lines of about 80 grid lines/em (NCRP, 1986). Moving grids arethinner than conventional Bucky' grids and have carbon fibre interspace material. Mostmodem mammographic X-ray units are designed with moving grids to blur the grid linesthereby reducing its presence in the image. A moving grid is essential to good imagequality despite the associated increase patient dose. The increase in patient radiation doseassociated with the use ofgrids may be offset by: (i) using higher tube voltage settings, (ii)increase in X-ray beam filtration and (iii) using higher speed recording systems (NCRP,1986).1.2.2.6. MagnificationScreen-film mammography utilizes high speed screen-film systems which allow the use ofsmaller focal spots with corresponding improvement in image sharpness. Magnification is1 Bucky is a film cassette holder, which contains a moving grid.8

the enlargement ofsuspicious areas ofa mammogram! with the goal ofbetter visualizationof fine tissue structures and the provision of more diagnostic information. Magnificationmammography is another advantage ofscreen-film mammography.1.2.2.7. Film processorFilm processor is an integral part ofscreen-film mammography. The film processing is thestage ofthe mammographic technique whereby the visible image captured permanently onthe X-ray film is developed to produce an image of the breast area under examination. Thefilm processor unit consists of developer, a fixer and water. The developer, a reducingagent or a donor ofelectrons donates electrons to the silver cations to reduce it to neutralsilver atoms; the fixer, usually sodium or ammonium thiosulphate, removes undevelopedsilver iodobromide crystals still present on the film by binding tightly with the remainingsilver ions to fonn a water-soluble complex; the processed film is finally washed in waterand dried (Wolbarst, 1993). It is important that the film processor is kept in good conditionfor good and quality mammograms.The capacity of a mammographic image to convey clinically useful information is measuredby three especially important parameters namely; contrast, resolution, and noise level. Sincethe production of high-contrast, high-resolution and low noise images with the lowestradiation dose possible is the goal of mammography, there have been several investigationscarried out for the optimisation and improvement of mammography techniques (Fahrig andYaffe, 1994a, 1994b; Lado et a!., 1997; Bhat et a!., 1998a, 1998b; Anastasio et al., 1998; andKallergi et a!., 1999). Dose and image quality in mammography studies conducted by Young( A mammogram is the mammographic X-rayimage ofthe breast developed on an x-ray film.9

et al. (1996) concluded that for all breast thicknesses, highest contrast and therefore overallimage quality is achieved using conventional mammography techniques. Review ofmammography techniques and evaluation of real cost and benefit ratios by Simonetti et al. in1998 also concluded that conventional mammography remains the most sensitive tool forbreast cancer diagnosis. An X-ray beam from a screen-film mammography unit is used in thisstudy.1.3. Interactions of X-rays with matterThe interaction of X-rays or photons with matter results in a number of possible interactionmechanisms such as: photoelectric absorption, coherent or Rayleigh scattering, incoherent orCompton scattering, pair production and photonuclear interactions. Photoelectric absorption,Compton scattering and pair production are the most important interactions in radiation dosemeasurements as they lead to partial or complete transfer of photon energy to electron energywhich consequently impart energy into matter. Being an elastic scattering interaction,Rayleigh scattering involves the redirection of the photon through a small angle with noenergy loss. Photonuclear interactions become significant when photon energy is in excess ofa few MeV (Attix, 1986). The kinematics and the interaction probabilities or interactioncoefficients involve and depend largely on the energy of the incident photon and the physicalproperties ofthe target material such as atomic number, Z, and density.1.4. Mammography DosimetryThere is a small but non-negligible risk of radiation-induced carcinogenesis associated with anX-ray examination of the female breast (NCRP, 1986). In fact, in the mid 1970s it waspostulated in the U.S. that the National Cancer Institute's screening mammography programIO

induced more cancers than were found (Bailer, 1977). Although more careful analysis of theradiation dose and associated risks estimates has negated this hypothesis, mammographyradiation doses are of concern and are routinely monitored. The magnitude of the absorbedradiation dose to the breast from mammography X-ray beams forms an important part ofthequality control of the mammographic examination since it gives an indication of theperformance of the mammographic imaging system as well as an estimated risk to the patient.Breast cancer almost always arises in the glandular tissue of the breast. As a result, the meanor average radiation absorbed dose of the glandular tissue is the preferred measure of theradiation risk associated with mammography (NCRP, 1986; Rosenstein et a!., 1985). Themean glandular (MGD) dose is the quantity also recommended by International Commissionon Radiological Protection (ICRP, 1991) and is used by many national protocols, such as theEuropean Protocol (CEC, 1996).Mammographic dosimetry is primarily to assess the risk of radiation-induced carcinogenesisin mammographic examinations. Breast dose assessment has therefore been recommended tobe included in every mammographic quality assurance programme by some national protocolsand institutions such as the European Protocol, using the mean glandular dose (MGD) as therisk assessment parameter since the glandular tissue is the most vulnerable of the breast tissues(European Protocol (CEC, 1996), the British Institute of Physical Sciences in Medicine(IPSM, 1989), and the International Commission on Radiological Protection (ICRP, 1991)).Measurements of the mean glandular dose from mammography have been carried out by anumber ofinvestigators using a variety ofmammographic techniques.II

An accurate knowledge of the output of an X-ray tube is essential in medical diagnostics forensuring accurate dose levels estimation and for the provision of a useful check on thediagnostic adequacy of the technique. Increased use of mammography has brought into focusthe necessity for radiation dose reduction. The application of X-rays and ionizing radiationsfor diagnostic radiology requires that the procedure is justified and optimized and that theexposure to the patient is kept as low as possible, without compromising image information.1.5. Introductory Description to StudiesThis thesis describes a system that will provide instantaneous radiation information frompatients undergoing mammography examination. Standard conventional techniques for themeasurement ofMGD which involves measurement of tube loading (TL) and entrance surfaceair kerma (ESAK) for dose calculations were employed to calculate radiation dose frommammography X-ray beams. Other methodologies also presented include: (1) directmeasurement ofX-ray spectra using a germanium detector and computation ofmean glandulardose from the measured spectra; (2) computation ofMGD from theoretically generated X-rayspectrum using a Monte Carlo simulation code for electron and photon transport, PENELOPE.This work also focused on exploring the suitability of diamond as a detection system formammography dosimetric application. Both single crystal diamond produced under highpressurehigh-temperature (HPHT) conditions and polycrystalline diamond produced bychemical vapour deposition (CVD) were used in the study. The tissue equivalence, chemicalinertness, basic material and electrical properties as well as the radiation sensitivity ofdiamond are important parameters in this study. Investigation into the mechanisms necessaryto enhance the absorption capabilities of the diamond films was undertaken. A suitable12

diamond probe has been constructed for measuring integrated dose from mammography X-raybeams.1.5.1. Overview ofDirect Measurement of Mean Glandular DoseThe standard method of estimating the mean glandular dose on patients undergoingmammography X-ray examinations is based on incident air kerma or entrance surface airkerma (ESAK) measurements without backscatter and the conversion to glandular dose usingappropriate conversion factors depending on the type of phantom used (IPSM, 1994). The airkerma value may be determined either for patients or for a standard breast phantom (Stanton etal., 1984). Polymethyl methacrylate' (PMMA) also known as acrylite, Perspex, Plexiglas andLucite are normally used as breast substitute phantoms. The incident air kerma is obtainedfrom the product of (1) the exposure time current product (mAs) i.e. tube loading for correctexposure (recommended optical density) of the PMMA phantom and (2) the output (airkermaper mAs) of the X-ray machine with the phantom removed. The mean glandular dose is thenobtained either by the relationshipD=Kpg.K is the incident air kerma without backscatter at the specified halfvalue layer; g is the ESAKto MGD conversion factor and the p-factor converts air kerma for the PMMA breast substitutephantom to that for the model breast (Dance et a!., 1999). A direct radiation dose assessmentcan also be made with thermoluminescent dosemeters (TLDs) placed on the breast toI PMMA is normally used as a breast substitute phantom. in view of its noble properties such as: (i) high lighttransmittance with a refractive index of 1.49; (ii) non conducting with high electrical resistance; (iii) unaffectedby prolonged exposure to moisture; (iv) can easily be heat-molded without loss of optical clarity; (v) highstrength-to-weight ration; (vi) can easily be sawed, drilled, milled and engraved (Boedeker, 200~; MatWeb,2004). pM1>IA is also cheap and readily available.13

determine the dose (K,) to the entrance surface of the breast (this will include all backscatteredradiation). The MGD has also been calculated through the use of energy fluence or x-rayenergy spectra distribution (Boag et al., 1976; Shrivastava, 1981; Skubic and Fatouros, 1986;Calicchia et al. (1996) or computed by using Monte Carlo techniques (Dance, 1980 andRosenstein et al., 1985). In Chapter 2 of the thesis, the results of the MGD calculations arepresented. Part of the MGD measurements has been published and the article presented inChapter 2 (Assiamah et aI., 2004).1.5.2. Overview of Direct X-Ray Spectra MeasurementAn accurate knowledge of the output of an X-ray tube is essential in many areas of study. Itforms the basis ofalmost all image quality simulation. Direct measurement ofX-ray spectra isusually performed with a high-purity germanium detector with the signal output beingprocessed by conventional electronics and a multichannel analyzer (MCA). Notable amongthese works include, 'Fewell and Shuping (1977), Birch and Marshall (1979), Fewell et al.(1981). Others are Aoki and Koyama (1989), Laitano et al. (1991), and Marshall et al. (1996).The relatively complex nature of X-ray spectral measurement from diagnostic X-ray machinesmakes direct measurement ofX-ray spectra a difficult task. High photon fluxes produced fromdiagnostic X-ray machines and the consequent probability of pulse pileup occurring, hencedistorting the detector output are the main difficulties in the measurement of X-ray spectrafrom a mammography X-ray machine. Additional problems are:- (i) the relatively low energyof X-rays generated by a mammography machine. These photons can be significantlyattenuated by small amounts ofany medium including air. (ii) The unavailability ofthe optionto reduce the photon flux emitted from the X-ray tube by using a low current or fluoroscopic14

mode on a typical mammography machine compounds the complex nature of direct X-raymeasurement from an X-ray machine.Several solutions have been found to address the problem of pile-up effects. They include, theuse of an appropriate gating signal to reduce pulse pileup, the use of an air-free path betweenthe X-ray machine and detector and the use of multi-parameter system (MPS) instead of astandard multi-channel analyzer (MCA). Application of these novel solutions by Wilkinson etal. (2001) to address the high flux production and associated pulse pileup indicate that usingMCA or MPS with an air-free path in conjunction with a gating signal is an effectivetechnique to solve the problem. However, an air-free path is difficult to implement and so isunsuitable in a clinical setting where a simple, convenient method is desirable. Wilkinson etal. have suggested the use of an MCA with a gating technique. They however, noted thatneither of the techniques generates an absolute measurement of the photon flux since thepileup rejection process rejects some detected events.In this thesis, the method ofdata acquisition, adoption and optimization ofexisting proceduresfor correcting or stripping spectral data with the aim of placing the incident photon in itsprimary and correct energy bin has been carried out. The corrected photon flux has been usedto calculate mean glandular dose employing appropriate conversion factors. Mass attenuationcoefficient data and mass energy-absorption coefficient data for certain materials wereemployed in various stages of the spectrum corrections. The attenuation coefficient data wereobtained from the published results of Hubbell and Seltzer (1996). The mass attenuation andmass energy-absorption coefficient was fitted to an extension of a fifth degree polynomial15

equation (Tucker et aI., 1991) in order to derive the parameters needed to generate theabsorption energy "spectra". This was then used in computing the air kerma K (E) andconsequently, the MGD. In Chapter 2 the outcome of the spectral measurement andconsequent computation of mean glandular dose is given. The extension to the Tucker et al.equation is found necessary, as the published data do not cover the full energy spectrum ofinterest. In Chapter 3 is presented the steps used to generate the absorption energy "spectra"from existing published data. The effect ofambient temperature, pressure and humidity on thecalculated values ofphotons and air kerma were also considered and the methods involved arepresented in Chapter 4 ofthe thesis.1.5.3. Overview ofTheoretical Computations ofX-Ray SpectraThe difficulties associated with direct measurement of X-ray spectra led many researchers todevelop mathematical models the use of which enables the calculation of a spectrum for adefined peak kilovoltage and X-ray target/filter combinations to generate X-ray spectrarequired. The earliest spectra calculations are those of Kulenkampff in 1922 and Kramers in1923. There were two shortcomings in the Kramers model. The first was the assumption thatthe original differential energy intensity was constant. The second drawback was that theabsorption of the X-rays in the target was not accounted for. Krarners' model has beencorrected and modified by later researchers; Unsworth and Greening, (1970); Sundararaman etaI., (1973); SooIe (1977); and Birch and Marshall (1979). There has been tremendousdevelopment and advancement in the generation ofX-ray spectra by mathematical models andcalculations. Computer simulation has become a convenient and frequently used tool in thestudy of X-ray mammography for the calculation of radiation doses to the breast (Wu et aI.,16

1991); for optimizing techniques in mammography (Fahrig and Yaffe, I994a, I994b;Sandborg et aI., 1994); and for evaluating detector performance (Chen, 1980; Maidment et aI.,1993). Computer-generated spectra in mammography enables system designers to predictpatient dose more accurately and, hence, aid in the development of better hardware andsoftware systems to reduce patient dose. In recent times, several models have been proposed;they include, the Tucker et aI., (199Ia, 199Ib); Boone and Seibert, (1997); Boone et al.(1997); and Blough et aI., (1998) models. The most popular of these models formammography has been that ofTucker et al. Their calculations attempt to model the physicalprocesses such as self-absorption within the target, and the use of a semi-empirical approachin fitting a parameterized function to previously measured spectra. In their comparison ofmammography spectral measurements with spectra produced using several differentmathematical models, Wilkinson et aI., (2001), have indicated that accurate results can beproduced by all mathematical models, but only ifthe user attempts to match the calculated halfvalue layer (HVL) of the modelled spectrum with the physically measured HVL. Themodelled spectra may otherwise be in error and can lead to an underestimation of the dosecalculation by up to 20%.In this thesis a computer code system PENELOPE (PENetration and Energy Loss of Positronand Electrons) which simulates coupled electron-photon transport has been studied and used tocompute radiation dose from mammography X-ray beam. The simulation algorithm is basedon a scattering model that combines numerical databases with analytical cross section modelsfor different interaction mechanisms (Salvat et aI., 200 I). A model was developed that17

simulates the experimental setup and conditions for evaluating manunography X-ray beamradiation dose. Presented in Chapter 6 is the simulation work using the PENELOPE MC code.1.5.4. Overview of Dosimetry with DiamondDiamond is a unique element with attractive physical properties such as wide band gapmaking it an excellent insulator, wide transmission band, optically transparent, high chargecarrier mobility, high breakdown voltage, high thermal conductivity and low thermalexpansion coefficient, small dielectric constant, excellent radiation hardness, physically hardand chemically inert. The near tissue equivalence and chemical inertness of diamond havebeen identified to be important in both medical physics and health physics as a radiationmonitoring tool (Nam et al., 1987). Thus, as a radiation dosimeter, it does not require largecorrection factors to convert the response to a true deposited radiation dose. This is becausethe energy response or energy absorption of diamond as a dose-sensing element per unitexposure will be similar to that oftissue.Diamond also has low noise contribution arising from leakage currents due to the large bandgap. Further, good sensitivity compared to ionization chambers has been found (Bruzzi et aI.,2000) and the insensitivity to radiation damage makes diamond a material more suitable thanmost solid-state detectors currently being used for on-line dosimetric applications (Plankskoy,1980; Rustgi, 1995).The applications of synthetic diamond crystal as radiation sensing elements have beenconcentrated mainly on its response to charge particles. These applications have been both innuclear and medical physics. Some ofthe studies include: (Burgemeister, 198I) rate ionization18

chamber radiation detectors in biological environments (Keddy et aI., 1987), as athermoluminescent detector (Nam, 1989), as a sensor for measuring low dose-rates(Grobbelaar et aI., 1991), as a near tissue-equivalent probe in electron radiation therapy (vander Merwe, 1994), as detectors in the heavy-ion dosimetry for tumor-therapy (Berdermann etaI., 1998). There are also extensive reports on the application of synthetic diamond asradiation detectors in high energy physics studies (Meier, 1999; Adam et aI., 2000; Adam etaI., 2002; Berdermann et aI., 200 I; Shu et aI., 200 I; Bergonzo et aI., 2003) due to the radiationhardness ofdiamond which results from the high energy (80 eV) needed to remove a carbonatom from the diamond lattice and the high thermal conductivity which is about five timeshigher than for copper.However, its use for both low energy X-ray measurements and dose determination has notbeen fully investigated. Diamond has an advantage because of the lower energy required toproduce a charge pair, and the fact that diamond can be used in smaller sizes than gas-filledionization chambers, also permitting a better spatial resolution for measurements of radiationfields with steep dose gradient. The project aim was to research and utilize such keycharacteristics of diamond. Diamond has been used in a conduction mode whereby electrodesare connected to it and the resulting current from the interaction of the ionizing radiation withthe diamond was detected. A new method for calculating total dose to tissue has beenestablished from the studies. The studies on diamond as a radiation detector for mammographybeam dosimetry are present Chapter 5. The general conclusions from each of the techniquesused in this study is consolidated and reported in Chapter 7.19

REFERENCESAdam W. et al. (RD42 Collaboration) (2000). Pulse height distribution and radiation toleranceofCVD diamond detectors. Nuc!. Instrum. Methods Phys. Res. A 447,244-250.Adam W. et al. (RD42 Collaboration) (2002). Radiation tolerance ofCVD diamonddetectors for pions and protons. Nuc!. Instrum. Methods Phys. Res. A 476,686-693.American Cancer Society (200 I). Surveillance research: breast cancer, facts and figures.Anastasio M.A., Yoshida H., Nagel R., Nishikawa R.M., and Doi K. (September 1998). Med.Phys. 25 (9),1613.Aoki K. and Koyama M. (1989). Measurement of diagnostic X-ray spectra using a siliconphotodiode. Med Phys. 16, 529-536.Assiamah M., Nam T.L., and Keddy R.J. (2004). Dosimetric techniques for mammographyX-ray beams. Proceedings of the 9 th International Symposium of Radiation Physicsand Workshop on Radiation Based Analytical Techniques, 24-31 October 2003, CapeTown, South Africa. Radial. Phys. Chern. 71, 957-958.Attix F.H. (1986). Introduction to radiological physics and radiation dosimetry. Wiley­Interscience, USA, 154.Bailar J.e. (1977). Cancer 39,2783-2795.Bames G.T. and. Frey G.D (editors) (1991). Screen-Film Mammography: ImagingConsiderationa and Medical Physcis Responsibilities. Proc. Of SEAAPM SpringSymposium, Columbia, South Carolina, April 6, 1-13 and 119.Berdermann E. et al. (RD42 Collaboration) (200 I). The use of CVD diamond for heavy-iondetection. Diamond and Related Materials 10, 1770-1777.20

Boone J.M., Fewell T.R, and Jennings R.J. (1991). Molybdenum, rhodium, and tungstenanode spectral models using interpolating polynomials with application tomammography. Med. Phys. 24 (12), 1863-1814.Bruzzi M., Bucciolini M., Cirrone G.A.P., Cuttone G., Mazzocchi S., Pirollo S., and SciortinoS. (2000). Characterization of CVD diamond dosimeters in on-line configuration.Nuc!.lnstrum. Methods Phys. Res. A 454,142-146.Burgemeister E.A. (1981). Dosimetry with a diamond operating as a resistor. Phys. Med. Bio!'26 (2), 269-215.Chen C.S., Doi K., Vybomy c., Chan H.P., and Holje G. (1980). Monte Carlo simulationstudies ofdetectors used in the measurements of diagnostic X-ray spectra. Med. Phys.1,621-635.Dance D. R., Skinner C. L., and Carlsson G. Aim (1999). Breast dosimetry. Appl. Radial. Iso.50, 185-203.European Protocol (CEC) (1996). European protocol on dosimetry III mammography.European Commission, Luxembourg, (EUR 16263).Fahrig R. and Yaffe M.J. (1994a). A model for optimization ofspectral shape in digitalmammography. Med. Phys. 21 (9),1463-1411.Fahrig R. and Yaffe M.J. (l994b). Optimization ofspectral shape in digital mammography:dependence on anode material, breast thickness, and lesion type. Med. Phys. 21 (9),1473-1481.Fewell T.R. and Shuping R.E. (1911). Photon energy distribution of some typical diagoosticX-ray beams. Med. Phys. 4,181-196.Fewell T.R., Shuping R.E., and Healy K.E. (1981). Handbook ofcomputed tomography X-rayspectra. HHS Publication (FDA) 81-8162, Washington, D.C.22

GLOBOCAN (2000). Cancer incidence, mortality and prevalence worldwide, version 1.0,IARC Cancer Base No.5. Lyon, IARC Press.Grobbelaar I.H.,Bums R.C., Nam T.L and Keddy R.J. (1991). Miniaturized radiationdetector with custom synthesized diamond crystal as sensor. Nuc!. Instrum. Methods.Phys. Res. B61, 553-559.Hubbell J. H. and Seltzer S. M., 1996. Tables of X-ray mass attenuation and mass energyabsorption coefficients from I keV to 20 MeV for elements Z= I to 92 and 48additional substances of dosimetric interest. National Institute of Standards andTechnology, U.S. Department ofCommerce, Gaithersburg, MD 20899.International Commission on Radiological Protection (ICRP) (1991). ICRP Publication 60,Annals of the ICRP 21, 1-3.Institute of Physical Sciences in Medicine (IPSM) (1989). The Commissioning and RoutineTesting ofMarnmographic X-ray systems. IPSM Topical Group Report 59.Institute of Physical Sciences in Medicine (IPSM) (1994). The Commissioning and RoutineTesting of Mammographic X-ray systems. 2 odedition. IPSM Topical Group Report59.Kallergi M., Carney G.M., and. Gaviria J (February 1999). Evaluating the performance ofdetection algorithms in digital mammography. Med. Phys. 26 (2), 267-257.Keddy RJ., Nam T.L., and Bums R.C. (1987). Synthetic diamonds as ionization chamberradiation detectors in biological environments. Phys. Med. Bio!. 32 (6), 751-759.Kramers H.A. (1923). On the theory of X-ray absorption and of the continuous X-rayspectrum. Philo. Mag. 36, 836-87 I.KulenkampffH. (1922). Uber das kontinuierliche Rontgenspektrum. Ann. Phys. (Leipzig) 69,548-596.23

Lado M.J., Tahoces P.G., Souto M., Mendez A.J., and Vidal J.J. (September 1997). Real andsimulated clustered microcalcifications in digital mammograms. ROC study ofobserver performance. Med. Phys. 24 (9), 1385-1394.Laitano R.F., Pani R., and Pellegrini R. (1991). Determination of X-ray spectra out ofscattered component up to 300kV. Med. Phys. 18,934-938.Lee L., Stickland V., Wilson Robin, and Roebuck E. (1995). Fundamentals ofmammography.W.B. Saunders, 2.Maidment A.D.A., Fahrig R., and Yaffe M.J. (1993). Dynamic range requirements in digitalmammography. Med. Phys. 20,1621-1633.Marshall N.W., Faulkner K., and Warren H (1996). Measured scattered X-ray energy spectrafor simulated irradiation geometries in diagnostic radiology. Med. Phys. 23, 1271­1276.MatWeb Online Material Data Sheet (2004). MatWeb Material Property Data.Meier D. (1999). CVD diamond sensors for particle detection and tracking. PhD thesis,University ofHeidelberg, Germany.Nam T.L., Keddy R.J. and Bums R.C. (1987). Synthetic diamonds as III VIVOradiationdetectors. Med. Phys. 14 (4), 596.Nam T.L. (1989). Nuclear radiation detection properties of diamond. PhD thesis, Faculty ofScience, University ofthe Witwatersrand, Johannesburg, South Africa. 72-77.National Cancer 1nstitute (1999). Screening for breast cancer. Electronic text redistributed byUniversity of Bonn, Medical Centre.National Council on Radiation Protection and Measurements, (NCRP) (1986). NCRPReport number 85. Bethesda, MD, 7-23, 40-48.24

Plankskoy B. (1980). Evaluation ofdiamond radiation dosemeters. Phys. Med. BioI. 25, 519.Rosenstein M., Anderson L. W., and Warner G. G. (1985), Handbook ofglandular tissuedoses in mammography. HHS Publication FDA 85-8239 (Centre for Devices andRadiological Health, Rockville, Maryland 20857, USA).Rothenberg L.N. and Hans A.G. (November 1995). Physicists in mammography - a historicalperspective. Med. Phys. 22 (11),1923-1934.Rustgi S.N. (1995). Evaluation of the dosimetric characteristics of a diamond detector forphoton beam measurements. Med. Phys. 22, 567.Salvat F., Fernandez-Varea J.M., Acosta E., and Sempau 1. (2001). PENELOPE - a codesystem for Monte Carlo simulation of electron and photon transport. (OECDINEAData Bank, Issy-ies Moulineaux, France, 2001). Available in PDF format fromwww.nea.fr.Sandborg M., Carlsson C.A., and Carlsson G.A. (1994). Shaping X-ray spectra with filters inX-ray diagnostics. Med. BioI. Eng. Cornput. 32, 384-390.Shrivastava P.N. (1981). Radiation dose in mammography: an energy-balance approach.Radiology 140,483-490.Shu D., Job P.K., and Kuzay T.M. (2001). CVD-diamond-based position-sensitivedetector test with electron beam from a Rhodotron'" accelerator. Proceedings of the2001 Particle Accelerator Conference, Chicago, 2435-2437.Siemens Medical Solutions Website (2004). http://www.med.siemens.com/Simonetti G., Cossu E., Montanaro M., Caschili c., and Giuliani V., (1998). What's new inmammography. European Journal ofRadiology 27, S234.Skubic S. E. and Fatouros P. P. (1986). Absorbed breast dose: dependence on radiographicmodality and technique, and breast thickness. Radiology 161,263-270.25

Soole B.W. (1977). A determination by an analysis of X-ray attenuation in aluminium of theintensity distribution at its point of origin in a thick tungsten target of bremsstrahlungexcited by constant potentials of 60-140 kV. Phys. Med. Biol. 22, 187-207.Stanton L., Villafana T, Day John L., and Lightfoot Davis A. (1984). Dosage evaluation inmammography. Radiology 150,577-584.Sundararaman V., Prasad, M.A. and Vora R.B. (1973). Computed spectra from diagnostic andtherapeutic X-ray tubes. Phys. Med. BioI. 18(2),208-218.Tucker D.M., Barnes GT., and Chakraborty D.P. (1991a). Semi-empirical model forgenerating tungsten target X-ray spectra. Med. Phys. 18, 211-218.Tucker D.M., Bames G.T., and Wu X. (199Ib). Molybdenum target X-ray spectra: a semiempiricalmodel. Med. Phys. 18,402-407.Unsworth M.H. and GreeningJ.R. (1970). Theoretical continuous and L-characteristic X-rayfor tungsten target tubes operated at 10-50kVp. Phys. Med. BioI. 15,621-630.van der Merwe D.G. (1994). The effect oftissue inhomogeneities on the energy spectrumand dosimetry in electron radiation therapy. PhD thesis, Faculty of Science,University ofthe Witwatersrand, Johannesburg, South Africa.Wilkinson L. E., Johnston P _N., and Heggie J. C. P. (200 I). A comparison of mammographyspectral measurements with spectra produced using several different mathematicalmodels. Phys. in Med. Biol. 46,1575-1589.Wolbarst A. B. (1993). Physics of Radiology. Int. ed., Appleton and Lange, Connecticut06855 USA, 122-131,140.Wu X., Barnes G.T., and Tucker D.M. (1991). Spectral dependence of glandular tissue dose inscreen-film mammography. Radiology 179, 143-148.26

Chapter 2COMPARISON OF MAMMOGRAPHYRADIATION DOSE VALUES OBTAINED FROMDIRECT INCIDENT AIR KERMAMEASUREMENTS WITH VALUES FROMMEASURED X-RAY SPECTRAL DATAAccepted for publication in Applied Radiation and Isotopes28

AbstractThe application of X-rays and ionising radiations for diagnostic radiology requires that theprocedure is justified and optimised and that the exposure to the patient is kept as low aspossible, without compromising image information. X-ray mammography is considered to bethe most sensitive technique currently available for early detection of breast cancer. Themagnitude of the absorbed radiation dose to the breast from mammography X-ray beamsforms an important part ofthe quality control ofthe mammographic examination since it givesan indication of the performance of the mammographic imaging system as well as anestimated risk to the patient. In this work mean glandular dose (MGD) values were obtained atvarious tube potentials and tube loadings (TL) using direct measurements of the incident airkerma (ESAK) at the surface of a standard breast phantom and also from spectralmeasurements acquired with a solid-state detector. Comparisons of the MGD values thusderived are presented and the relationship between MGD, phantom thickness, image qualityand tube operating parameters is discussed.2.1. IntroductionBreast cancer is reported to be the second highest cause of cancer deaths in women today(after lung cancer), with an estimation of 400,600 deaths from breast cancer for 2001(GLOBOCAN, 2000). Breast cancer is also the most common cancer among women,excluding nonmelanoma skin cancers. In 2001, World Health Organization (WHO) predictedmore than 1.2 million new breast cancer cases (globally) while in United States, both theAmerican Cancer Society and the National Cancer Institute estimated approximately 192,200new cases ofinvasive breast cancer cases for 2001. Male breast cancer has also been reported29

ut this accounts for less than 1% of the total (American Cancer Society, 2001; NationalCancer Institute, 1999). Statistics on mammography indicate that the incidence of breastcancers per 100,000 women increased by approximately 4% during the 1980s and levelled offto about 100 cases per 100,000 women in the 1990s. The reported death rates from breastcancer also declined significantly between 1992 and 1996. Medical experts attribute thereduction in breast cancer death to earlier detection and more effective treatment.The use of X-ray mammography in the screening of asymptomatic women has become verycommon in many parts of the world. The objective of the screening exercise is to detect breastcancer at an early stage to reduce breast cancer mortality. There is however, a small but nonnegligiblerisk of radiation-induced carcinogenesis associated with an X-ray examination ofthe female breast (NCRP, 1986). An accurate knowledge of the output of an X-raymammography tube is essential in medical diagnostics for ensuring accurate dose levelestimation and for the provision ofa useful check on the diagnostic adequacy of the imagingtechnique. Breast radiation dose assessments are therefore included in most national qualityassurance programmes for X-ray mammography. Notable among the national protocols andinstitutions, which require this inclusion, are: NCRP (1986), AAPM (1990) and CEC (1996).The glandular tissue is the most vulnerable of the tissues making up the breast. Thus, amongthe different dosimetric quantities used in risk assessment, the mean glandular dose (MGD) isthe best indicator of the patient risk (NCRP, 1986) and it has been accepted by the nationalprotocols and institutions mentioned earlier and many others as the preferred quantity for themeasur,: of potential risk from mammography. Several other institutions have also suggested30

MGD as the preferred quantity for the measure of potential risk from mammography (ICRP,1987; NCRP, 1986; IPSM, 1989; IPSM, 1994). Measurements of absorbed radiation dosefrom mammography have been carried out by a number of investigators using a variety ofmammographic techniques (Young et a!., 1996; Klein et a!., 1997 and Dance et a!., 1999). Thestandard method of estimating the MGD dose on patients undergoing mammography X-rayexaminations is based on ESAK measurements without backscatter and the conversion toglandular dose using appropriate conversion factors depending on the type of phantom used(IPSM, 1994). The air kerma value may be determined either for patients or for a standardbreast phantom; polymethyl methacrylate (PMMA) is normally used as breast substitutephantom. The MGD is then obtained either by determining the tube loading or from theexposure parameters required to obtain the recommended optical density of the exposedstandard phantom. Two possibilities exist; either (i) the tube output at this tube loading ismeasured or, (ii) a direct radiation dose assessment using thermoluminescent dosemeters(TLDs) placed on the breast can determine the dose (K,) to the entrance surface of the breast(this will include all backscattered radiation).Spectral information is essential for dosimetric purposes. An accurate knowledge of the X-rayspectral output forms the basis of almost all image quality simulations thereby enablingsystem designers to predict patient dose more accurately. Direct measurement of X-ray spectrais usually performed with a high-purity germanium detector with the signal output beingprocessed by conventional electronics and a multichannel analyzer (MCA) to obtain thephoton energy distribution. Spectrum correction techniques are then applied to the acquiredphoton distribution to generate the photon fluence spectrum from which the MGD is31

calculated (Calicchia et al., 1996). MGD values have also been calculated using Monte Carlotechniques (Wu et al., 1991).Unlike previously published works, this contribution provides an evaluation and comparisonof the values for the important MGD parameter as obtained by the two methods outlinedabove; the standard incident air kerma direct method and the spectral method that utilises theenergy fluence spectral data incident on standard phantoms of varying thicknesses. X-rayspectra were determined for a molybdenum anode with molybdenum filter at settings andexposure parameters normally used in routine mammography. The study also investigated therelationships between MGD, phantom thickness, image quality and tube voltages. Theirimplications in mammography are discussed.2.2.2.2.1.ExperimentalMaterialsThe X-ray unit used for this work was a General Electric (Model No. 65447) three-phaseSenographe SOOT six-pulse X-ray generator. The rotating anode had a grounded molybdenumtarget with nominal, selectable focal spot sizes of 0.3 and 0.1 mm. The X-ray unit has aninherent filtration of0.8 mm beryllium (an equivalent aluminium filtration of0.028 mm AI) at30 kVp and 0.03 mm molybdenum of added filtration. The breast substitute material used forthe study was PTW (Physikalisch Technische Werkstatten) acrylic glass, SIB MammographicPhantom (Type 42001). This phantom has attenuation properties similar to those ofPMMA ofsimilar thickness. The film and screen type used were AGFA Mamoray HDR (High DynamicRange) and AGFA Mamoray (High Dynamic Screen) respectively. For film processing, an32

AGFA Imaging MinlMed unit maintained at a temperature of 34.9°C with a processing cycleof 90 seconds was used. Film optical densities (00) were measured with X-Rite (Model no:331) densitometer. The densitometer has an accuracy of± 0.02 00 and reproducibility of±0.01 00. The X-ray tube output for the different exposure parameters was measured with aPTW-Diados mammography detector (Type T60005 0735) and PTW-Diados diagnosticdosemeter i.e. electrometer connected to a display unit (Type 11003-1129). The dosemeter anddetector were calibrated by PTW at energy values 25 - 45 kVp with a precision of less than0.5%. The calibration factor (NK) for each radiation quality was 1. The detector showed a flatenergy response over the mammography energy range and the dosemeter has reproducibilitybetter than ± 0.5% and a digital resolution of I nGy/s. A Gammex RMI Type 1100, 99.8%aluminium attenuator set was used for the attenuation and HVL measurements. Thealuminium attenuators were plates ofarea 10 cm 2 and ofthicknesses accurate to within ± 5%.For the spectral analysis, a high purity intrinsic planar, liquid nitrogen cooled, germaniumdetector coupled to multi-channel analyser (MCA) manufactured by DSG Detector Systems(Model No. PGP 200-10) was used. The detector was operated at a bias voltage of-2500 V.The detector had a thickness of 10 mm, a 0.015 mm thick beryllium window and 200-mm 2 ofsensitive area. The detector was coupled to a desktop computer fitted with Aptec SupervisorAutomation software cards for data acquisition and processing.2.2.2. Calibration of tube outputThe PTW-Diados mammography-energy calibrated detector was supported in the X-ray beamat 5 em above the cassette holder, 6 em from the chest wall edge and centred laterally. A tube33

current-exposure time product (tube loading) of40 mAs was selected. For this study, the tubeoutput was measured for tube voltages set at 25, 26, 27, 28, 29,30 and 32 kVp. CorrespondingHVL values, which were measured in the presence of the breast compression plate, are shownin Table 1. The measurements were all carried out with the breast compression plate in place.The tube output measurement has been expressed as air kerma per tube loading (~Gy/rnAs) atI metre focus-detector distance.2.2.3. Determination of tube loadingThe tube loading (TL) measurement for ensuring the correct exposure of the standard phantomwas determined following the procedure described by the European Commission (CEC, 1996).In this determination, the X-ray machine was set up for a cranio-caudal view, with the breastcompression plate and a loaded cassette in place. The phantom was placed on the breastsupport table for a cranio-caudal position. The compression plate was brought down onto thephantom. The phantom was exposed using exposure parameters employed clinically for astandard-sized breast and the tube loading was recorded. The exposed film was processed witha dedicated mammography processor and the optical density (aD) measured. An Ol)including base and fog was then verified (recommended Of) range given by the EuropeanProtocol i.e. 1.0 to 1.5 Of), (CEC, 1996)). The exposure settings were adjusted wherenecessary, and the procedure repeated until an Ol) ofapproximately 1.3 was achieved.34

2.2.4. Measurement of X-ray tube output ESAK per tube loadingThe PTW-Diados mammography detector was positioned at the reference point (phantomposition) during the tube loading determinations above the cassette holder for the 20, 30, 40, 50and 60 mm thicknesses of PMMA and 60 rom from the chest wall edge of the cassette holder.The radiation detector was exposed in the manual-mode, using the determined tube loadingsand the dosemeter or electrometer reading (air kenna) was recorded for each. Corrections toambient temperature, pressure and humidity values were made according to methods previouslydescribed (Assiamah et al. 2003a). The HVL for the selected tube loading was determined.2.2.5. Determination ofMGD from measured X-ray ESAKThe MGD was calculated from the ESAK (without backscatter), obtained from the outputmeasurements and standard conversion factor tabulations and the equation:MGD=Kgp, (2.1)where K is the ESAK at the specified HVL, g is the ESAK to MGD conversion factor. The p­factor converts air kerma for the PMMA breast substitute phantom to that for the model breastwhich it approximately simulates (Dance et al., 1999). Ambient temperature and pressurevalues were used to correct the dosemeter reading. All the MGD values presented are for afocal-spot to detector distance (FSD) of650 rom. The conversion factors g and p, used for thiswork were those from Dance (1990) and were calculated assuming the 'standard' breastphantom composition. The g factors for the measured half value layers were derived byinterpolation from the compiled data and gave a percent deviation ofless than 5%.35

2.2.6. Contrast measurementsFilm contrasts were carried out using PMMA thicknesses and experimental alignments similarto those used for the tube loading measurements. A 0.2 mm thick aluminium sheet ofarea 10 x10 mrrr' was centred on top of the PMMA. Film contrast was then obtained from theradiograph as the OD difference between the image of the aluminium square and the adjacentbackground (Desponds et aI., 1994).2.2.7. Detector calibration and determination of detector responseThe detector was calibrated for energy scales, linearity checks, resolution and full energy peakefficiency using standard sources Am-241, Co-57 and Ba-B3. An energy-channel conversionfactor of 40 eV per channel with 0 keY corresponding to 0 channel number was set for themeasurements. The detector system showed a linear energy response with a resolution of 500eV and 610 eV at 17.78 and 59.4 keY respectively.2.2.8. Spectral measurement and data correction techniquesFor a good geometry and also to reduce pile-up effects, a lead-copper collimator of size 2 emdiameter and ofthickness 1 mm each of lead and copper joined together was used to collimatethe X-ray beam. An X-ray photon distribution was generated with and without collimation foreach tube voltage. Three exposures were made at each tube setting. The measured photondistribution was corrected for detector efficiency and summed to obtain the number ofphotonsfor every 0.5 keY intervals. This keY interval was chosen to ensure that the K a and KfJ X-ray36

peaks appeared at the correct energy interval (17.5 and 19.5 keY respectively). The details ofthe spectra correction techniques are described in Mavunda et al, (2004).2.2.9. Determination ofMGD using spectral dataThe photon fluence (photons/mmi) obtained after correction for detector window and K-fluorescence escape was corrected for air attenuation as described by Assiamah et al. 2003a.The quality ofeach measured spectrum was modified by the addition ofaluminium filters untilthe I" HVL agreed with that of the I" HVL from the ionisation chamber measurement withbreast compression plate in place, but no added aluminium. The spectrum was normalized tocorrect for any variation caused by the addition of aluminium filters (Fewell, 1977). Todetermine the HVL that simulated the presence of the set up with the compression plate in thebeam path, the normalized photon energy fluence for each tube voltage was attenuated with 3rnm PMMA thickness (the equivalent thickness of the compression plate). Using thenormalized photon fluence remaining after PMMA attenuation¢i("0," ) ,exposure values,X,(E)r PH in roentgen (R) (Johns and Cunningham, 1983) at absolute temperature T,barometric pressure P and humidity H was calculated employing equation (2).(2.2)The subscript i refers to the i'h photon energy interval E, (keV),(f1,")~" is the energyabsorption coefficient of air at s.t.p.; PT.P.His the density of air at temperature T, pressure Pand humidity H (Cember, 1996). The exposure values calculated were converted to air kerma37

values using a factor 0.873 (Wolbarst, 1993). The MGD was calculated from the air kermavalues using Eqn. (I) and the appropriate conversion factors.2.2.10. Determination ofattenuation curve and halfvalue layerExposure values calculated as described previously were used to calculate attenuation pointsand the HVL of the spectral data. Points on the attenuation curve were found for eachthickness of aluminium attenuator by calculating exposure Xi (E) U.H and summing over allthe energy intervals. From the attenuation curve the HVL was calculated for each nominaltube voltage. To determine the HVL that simulated the presence of the compression plate inthe beam path, the normalized photon energy fluence for each tube voltage was attenuatedwith 3 mm PMMA thickness. The mass attenuation coefficient data, mass energy-absorptioncoefficient data and the elemental composition ofthe PMMA phantom used for this work wereobtained from the published results of Hubbell and Seltzer (1996). To facilitate thecomputation of the air kerma value, the mass energy-absorption coefficient i»;):"data byHubbell and Seltzer, was fitted to an extension of a fifth degree polynomial equation (Tuckeret aI., 1991) in order to derive the parameters needed to generate the absorption energy"spectra". This was then used in computing K(E). The extension to the Tucker et aI. equationis found necessary as the published (J-l,J:" data do not cover the full energy spectrum ofinterest (Assiamah et aI., 2003b).38

2.3. RESULTS AND DISCUSSIONThe HVL values ofthe mammography X-ray machine used for this study are shown in Table2.1. The variation in the tube output (u.Gy/m.As) versus the nominal tube voltage (kVp) ispresented in Figure 2.1. The comparison ofESAK (without backscatter) and the correspondingMGD calculated from both the direct and spectral data, for all the PMMA phantomthicknesses obtained at nominal tube voltages 25, 26, 27, 28, 29, 30 and 32 kVp are presentedin Table 2.2. The measured and calculated attenuation curve measurements for the set tubevoltages 26 and 28 kVp are shown in Figures 2.2 and 2.3 respectively for collimated X-raybeams and in Figures 2.4 and 2.5 for uncollimated X-ray beams. Figure 2.6 shows the contrastvalues as a function ofphantom thickness and nominal tube voltages for an optical density of1.3. In Figure 2.7 is presented the film contrast at 28 kVp with varying tube loadings fordifferent PMMA phantom thicknesses for a range ofoptical densities. Figure 2.8 shows X-rayimages of a 30 rnm SIB phantom exposed at a nominal tube voltage of 28 kVp with tubeloadings of25 and 32 mAs respectively while Figure 2.9 presents X-ray images of40 rnm and30 mm SIB phantom exposed with identical X-ray tube parameters (nominal tube voltage of28 kVp and 40 mAs).Although the targeted OD set for this work was 1.3 including base and fog, attaining this ODvalue was not always feasible at certain exposure parameters due to restrictions caused by thedesign of the X-ray machine. The tube output (flGy/mAs) (Figure 2.1) values measured werefound to be linear with the tube voltage settings, indicating that the mammography X-raybeam used was both reliable and consistent. For all the tube settings, the maximum deviationofeach tube output from the mean was less than 5%.39

Table 2.1: The HVL values of the mammography X-ray machine measured in thepresence ofthe breast compression plate used for this study at the specified nominal tubevoltages.Nominal rube voltageHVL(kVp)(mmAl)25 0.32 ± 0.0226 0.33 ± 0.0227 0.34 ± 0.0228 0.35 ± 0.0229 0.36 ± 0.0230 0.37± 0.0232 0.39± 0.0270.,.------------------------,6000 50-c E>.o:s 40a.'5 o2~f- 3020lOL-----_------------------!242526272829Nominal tube voltage (kVp)30313233Figure 2.1: Tube output (~Gy/mAs) for the nominal tube, altages between 25 and 32 (kVp) used in this study.40

The values of the ESAK and MGD (Table 2.2) from the measured air kerma and measuredspectral data agreed generally to within ±5% at all the nominal tube voltages and for thephantom thicknesses studied. The percentage difference between the two techniques at fewinstances was about 10%. The generally good agreement between the results from the twotechniques can be attributed to the collimating system that was used to reduce scatter radiationincident onto the germanium detector during spectral acquisition. lt was observed in earlierstudies, that larger components of scattered radiation incident on the germanium detectorcould lead to the calculated ESAK and MGD from the spectral measurements to differ athigher tube voltages and larger phantom thicknesses by as much as 100% from those of thedirect measurements, even though the first HVL was the same (Assiamah et aI., 2004). For thesame phantom thickness, the different exposure parameters gave significantly different ESAKand MGD values with lower nominal tube voltages resulting in higher ESAK and MGD valueswhen compared with the corresponding higher nominal kVps even though the optical densitieswere close. Young et aI., (1996) made similar observation. The dose reduction in using ahigher nominal kVp for a set PMMA phantom thickness was significant.The extent of thereduction however, was found to be dependent on phantom thicknesses. In particular, usingnominal 26 kVp or 29 kVp instead of nominal 25 kVp for.20 mm PMMA phantom thickness,resulted in a MGD reduction ofabout 30%. The accompanying loss in contrast from using thehigher nominal kVp was 10% and 15% respectively. For a phantom thickness of 30 mm, theloss in contrast in using nominal 26 and 27 kVps was 3%. The observed dose reductions were,in these cases, 3% and 6% respectively (Table 2.2).Using nominal values of 28 and 29 kVpto expose a 30 mm phantom resulted in a reduction in dose of 11.5% and 20% respectively.The corresponding contrast reductions were 8 and 11% respectively. In the case ofthe 40 mm41

Table 2.2: Comparison of ESAK and MGD values in mGy calculated from measured airkenna values and from measured spectra at nominal tube voltages 25, 26, 27, 28, 29, 30 and32 kVp. All values are for a focal-spat-detector distance of650 mm.Nominal tube PMMA Direct measurements Measured spectravoltage thickness ESAK MGD ESAK MGD(kVp) (mm)(mGy) (mGy) (mGy) (mGy)25 20 1.63 ± 0.11 0.71 ± 0.14 1.6 ± 0.32 0.7±0.1426 1.16 ± 0.08 0.52 ± 0.10 1.l4± 0.23 0.51 ± 0.1129 1.14 ± 0.08 0.54 ± 0.11 1.21 ± 0.24 0.58 ± 0.1125 30 3.1±0.22 0.96±0.19 3.14±063 0.97±0.1926 2.94 ± 0.21 0.93 ± 0.19 2.68 ± 0.5 0.78±0.1627 2.71 ±0.19 0.9±0.18 2.66 ± 0.53 0.88 ± 0.1828 2.51±0.18 0.85 ± 0.17 2.57± 0.51 087±0.1829 2.34 ± 0.13 0.78 ± 0.12 2.52 ± 0.5 0.85± 0.1725 40 8.14± 0.57 1.91± OJ8 7.98 ± 1.6 1.87± 0.3726 6.02 ± 0.42 1.45 ± 0.29 5.95 ± 1.19 1.43± 0.2927 5.82 ± 0.41 1.47 ± 0.3 5.31 ± 1.06 1.34± 0.2728 5.23 ± 0.37 1.35 ± 0.27 5.05 ± 1.0 1.31 ± 0.2629 4.54 ± 0.32 1.l7± 0.23 5.05 ± 1.0 1.31 ± 0.2629 50 9.31± 0.65 1.91 ± 0.38 10 ±2 2.27± 0.4530 8.3 ± 0.58 1.8 ± 0.36 8.12:!: 1.62 1.76:!: 0.3532 6.41 ± 0.45 1.44:!: OJ 6.51 ± 1.3 1.48 ± 0.332 60 13.41 ± 0.94 2.62 ± 0.52 13.03 ± 2.61 2.57:!: 0.5142

PMMA phantom thickness, using nominal kVp values of27, 28 and 29 instead of nominal 25kVp resulted in a MGD reduction of about 24%, 29% and 40% respectively. The losses incontrast in using the higher nominal kVp were respectively 6%, 9% and 14%. The dosereduction from using nominal 30 and 32 kVp values instead ofnominal 29 kVp, for a PMMAphantom thickness of 50 mm was 6% and 25% with a contrast loss of 3% and 20%respectively. The 60 mm PMMA phantom had the lowest contrast and the highest MGD value.The low contrast value observed in larger phantom thickness was ascribed partly to the largervalue of tube loading needed to expose the phantom resulting in longer exposure times withconsequent reciprocity law failure l .The dose reduction associated with higher tube voltages is due to the fact that for the samephantom thickness, higher tube voltages contain higher energy X-rays that penetrate thephantom with less absorption. The lower tube voltages on the other hand contain, relatively, asignificant number of lower energy X-rays that are more easily absorbed by the phantom,hence higher dose values result. For all PMMA phantom thicknesses, it was generallyobserved that contrast decreases with increasing tube voltage and that higher contrast wasachieved only by using lower tube voltage.Linear regression fits were conducted on plots of the logarithm of the relative attenuationvalues with varying aluminium filter thicknesses, obtained with (Figures 2.2 and 2.3) andI To generate a stable latent image, sensitivity speck in a grain ofsilver halide must accumulate a "critical mass"of silver atoms. If exposure rate is too slow i.e. long exposure times and therefore low flux, some silver atomscould be freed and leak away from the sensitivity speck resulting in criticality not achieved. This behaviour isreferred to as reciprocity law failure. In the case of mammography for a given kVp and mAs, films becomelighter as a result oflong exposure times (Wolbarst, 1993).43

without collimation (Figures 2.4 and 2.5) for both the spectral and direct measurements.Deviations from the linear regression fits obtained from both the measured spectra and directmeasurements for two tube voltage settings were compared. Goodness-of- fit tests (Walpoleand Myers, 1993) at 95% confidence interval were conducted for the attenuation values. Theattenuation data at 26 and 28 kVp had degrees offreedom of 12 and 10 respectively. The chisquareddistribution

Figure 2.2: Comparison ofrelative attenuation curve calculated from measured spectrum ofacollimated X-ray beam with the one measured directly for a tube voltage of26 kVp.100 '!!:--------------------------,--;Di;:;~tme~~A Spe{;~I_~..:.~~~ment :10IIIIoL- ~___lo0.51.5 2 2.5 3 3.5Alumoium thickness (rrm)44.555.5Figure 2.3: Comparison ofrelative attenuation curve calculated from measured spectrum ofacollimated X-ray beam with the one measured directly for a tube voltage of28 kVp.45

100 r"--------------;:====:=:;__,I:#: Direct measurement:1/1 Spectral measurement'L..._~_10Il:"oo 0,5Figure2.4: Comparison of the relative attenuation curve calculatedfrom measured spectrum ofanuncollimatedX-raybeam with thatmeasured directly fora nominaltube voltage of26 kvp.100 "[---------------=::;;::::::====-1f,.!:~f-l'-r~l:I::I:Direct measurementA Spectral measurement'i10I,° o 0,5 1,5 2 2.5 3 3.5Alumniumthickness (rrrn)4 45 5 5,5Figure 2.5: Comparison ofthe relative attenuation curve calculatedfrommeasuredspectrum ofanuncollimated X-raybeam with that measured directly fora nominaltubevoltage of28 kVp.46

The significant reduction in deviations of the attenuation values of the spectral measurementsfrom linearity fits when the X-ray beams were collimated lends further support to theobservation that the spectral measurement method is sensitive to the presence of scatterradiation. The study has also shown that exposed film contrast is a function of tube voltagesand phantom thickness (Figure 2.6); contrast decreasing with increasing tube voltage andphantom thickness. Figure 2.7 shows the film contrast at a nominal 28 kVp with varying tubeloading for different PMMA phantom thicknesses for a range of optical densities. The opticaldensity versus tube loading graphs ofthinner phantoms had a larger gradient when comparedto the values obtained from the larger thickness, implying that the selection of tube current forthinner phantoms should be done cautiously to achieve the right optical density and a goodcontrast. X-ray images of 30 mm phantom exposed at nominal tube voltage 28 kVp with 25and 32 mAs (Figure 2.8) illustrates the effect oftube loading on contrast. For the same set tubevoltage and tube loading, increasing phantom thickness decreases contrast as seen from Figure2.9. This is because of the preferential removal of the lower energy X-rays with thickerphantoms and a consequent increase in the effective energy ofthe beam therefore.For similar tube voltage and phantom thickness up to 40 mm, it was found that for every 10rnm increase in the phantom thickness, twice the mAs was needed to attain similar contrast.For larger phantom thicknesses however, more than twice the mAs was required for every 10rnm increase in order to achieve similar contrast. The observation suggests that the relationbetween contrast and phantom thickness has a steeper gradient for smaller thicknesses than forlarger thicknesses.47

0.45[_20rnnPMMA--G-30mnPMMAO'i _40rrmPMMA -l(-:l!rnnPMMA:-60 rrmPMMA. "03C0u1~0.35025 L\TXi.02O.E !-----_------_-_----~--_-_I24 25 26 27 28 29 30 31 32 33 34N:Jrrinal tubevoltage (kVp)Figure 2.6: Contrast of 0.2 mm aluminium as a function of PtvWA phantom thickness and tube voltage.The exposure setting was selected to give an optical density of 1.3.0.'03503020. "02E 0oo.e_20rrmPMMAo.i_30rrmPMMA__411rrmPMMA._-5lIrrmPMMA0.05__60rrmPMMA00 50 "EOTubeloading (rr04.s)2

Figurez.S: X-ray images of30 mm SIB phantom exposed at 28 kVp, 25 mAs (left) and at 28 kv p. 32 mAs(nght).Figure 2.9: X-ray images of40 mrn (left ) and 30 mrn (right) SIB phantom exposed with identical X-raytube parameters (28 kvp, 40 mAs).49

2.5. CONCLUSIONThe study has shown that both the direct and spectral measurement can be adopted to estimatethe MGD. It can be concluded that the measured air kerma method can be as accurate as themeasured spectral data method. It should however be borne in mind that values from the latterare based on large number of theoretical! y generated attenuation coefficient data that wererequired for the calculation. Highlighted in the studies is the sensitivity of the spectralmeasurement method to the presence of scatter radiation and the need for beam collimationand detector shielding in order to reduce radiation scattered onto the detector whilst acquiringspectral data. It can be concluded from the results of the study that films with similar opticaldensity do not necessarily have similar contrast values. Optical density is also not a directreflection of the deposited dose. The MGD values revealed that while the optical densities forsome films were similar, their ESAK and corresponding MGD values were significantlydifferent. It can further be concluded from the study that although using a higher tube voltageon thinner phantom thickness results in a lower MGD, 'the effect on contrast is significantlydetrimental and cannot be ignored. For the same X-ray target material and same filter type,lower tube voltages should be used to expose thinner thicknesses and higher tube voltages forlarger thicknesses. It has been found that for thinner phantoms, the best range of opticaldensity is 0.9 -2.0 and the equivalent contrast values are 0.25 - 0.38. For the thickerphantoms, the best optical density range is between 0.8 -1.6 and the equivalent contrastvalues are 0.2 - 0.34. Optical densities outside these ranges could compromise contrast.50

REFERENCESAAPM (American Association of Physicists in Medicine) (1990). Equipment requirementsand quality control for mammography. AAPM Report No. 29,45-61.American Cancer Society (2001). Surveillance research: breast cancer, facts and figures.Assiamah M., Mavunda D., Nam T.L., and Keddy R.J. (2003a). Effect of pressure,temperature and humidity in air on photon tluence and air kerma values at lowphoton energies. Radial. Phys. Chern. 68 (5),707-720.Assiamah M., Mavunda D., Nam T.L., and Keddy R.J. (2003b). Segmented multi fit ofpolynomial function for mass attenuation and energy-absorption coefficient values.Radial. Phys. Chern. 67 (I), 1-6.Assiamah M., Nam T.L., and Keddy RJ. (2004). Dosimetric techniques for mammographyX-ray beams. Proceedings of the 9 th International Symposium of Radiation Physicsand Workshop on Radiation Based Analytical Techniques, 24-31 October 2003, CapeTown, South Africa. Radial. Phys. Chern. 71, 957-958.Calicchia A., Gambaccini M., Indovina P.L., MazzeifF., and Pugliant L. (I 996).Niobiomlmolybdenum K-edge filtration in mammography: contrast and doseevaluation. Phys. Med. BioI. 4 I, 1717-1726.CEC (European Commission) (I996). European protocol on dosimetry in mammography.European Commission, Luxembourg (EUR 16263).Cember H. (I996). Introduction to health physics (3'd ed.). McGraw-Hill, 169-179.Dance D. R. (1990). Monte Carlo calculation of conversion factors for the estimation ofmean glandular breast dose. Phys.Med. BioI. 35,1211-1219.51

Dance D. R., Skinner C. L., and Carlsson G. Aim. (1999). Breast dosimetry. Appl. Radial.Iso. 50, 185-203.Desponds L., Klausz R., and Heidsieck R. (1994). Image quality and glandular dose forrhodium K-edge filter in mammography: performance comparison of the rhodiumanode with molybdenum anode. Coolidge Sc. Review 1,7-12.Fewell T. R. and Shuping R. E. (1977). Photon energy distribution of some typicaldiagnostic X-ray beams. Med. Phys. 4, pp. 187-196.GLOBOCAN (2000). Cancer incidence, mortality and prevalence worldwide, version 1.0,IARC Cancer Base No.5. Lyon, IARC Press.Hubbell J. H. and Seltzer S. M. (1996). Tables of X-ray mass attenuation and mass energyabsorption coefficients from I keV to 20 MeV for elements Z= I to 92 and 48additional substances ofdosimetric interest. National lnstitute of Standards andTechnology, U.S. Department ofCommerce, Gaithersburg, MD 20899.ICRP (lnternational Commission on Radiological Protection) (1987). Publication 52. Annalsofthe ICRP, (ICRP 17).IPSM (Institute ofPhysical Sciences in Medicine) (1989). The Commissioning and RoutineTesting of Mammographic X-ray systems. IPSM Topical Group Report 59, IPSM,York.IPSM (lnstitute of Physical Sciences in Medicine) (1994). The Commissioning and RoutineTesting of Mammographic X-ray systems. IPSM Topical Group Report 59 (2'dedition), IPSM, York.52

Johns H. E. and Cunningham J. R. (1983). The physics of radiology (4 th edition). Thomas,Springfield, Illinois, 261-263.Klein R., Aichinger H., Dierker J., Jansen J. T. M., Joite-Barfub S., Sabel M., Schulz­Wendtland R., and ZoeteliefJ. (1997). Determination of average glandular dose withmodem mammography units for two large groups of patients. Phys. Med. Biol, 42,651-671.Mavunda R.D., Assiamah M., Narn T.L., and Keddy R.J. (2004). Bremsstrahlung spectra fromdiagnostic X-rays. Proceedings of the 9 th International Symposium of RadiationPhysics and Workshop on Radiation Based Analytical Techniques, Cape Town, SouthAfrica, 24-31 October 2003. Radial. Phys. Chern. 71, 991-992.NCRP (National Council on Radiation Protection and Measurements) (1986).Mammography- a user's guide. Bethesda, MD: National Council on RadiationProtection and Measurements, (NCRP Report 85), 40-56.National Cancer Institute (1999). Screening for breast cancer. Electronic text redistributedby University of Bonn, Medical Centre.Tucker D. M., Barnes G.T., and Charkraborty Dev P. (1991). Semi-empirical model forgenerating tungsten target X-ray spectra. Med. Phys. 18 (2),211-218.Walpole R.E. and Myers R.H. (1993). Probability and statistics for engineers and scientists(5 th edition), Prentice-Hall, Inc., Simon and Schuster Company, Englewood Cliffs,New Jersey 07632, 344-412.Wolbarst A. B. (1993). Physics of radiology (International edition). Appleton and Lange,Connecticut 06855 USA: 122-131, 185.53

Wu X., Barnes G.T., and Tucker D.M. (1991). Spectral dependence of glandular tissue dosein screen-film mammography. Radiology 179, 143-148.Young K.C., Ramsdale M.L., and Rust A. (1996). Dose and image quality in mammographywith an automatic beam quality system. The British Journal of Radiology 69,555-562.54

ELSEVIERAvailable online at www.sciencedireet.comSCIENCE@DIREcyeRadiation Physics and Chemistry 71 (2004) 957-958Radiation PhysicsandChemistrywww.e1sevier.com/1ocate/radphyschemDosimetric techniques for mammography X-ray beamsM. Assiamah*, T.L. Nam, RJ. KeddyHealth Physics Servicet Schonland Research Institute for Nuclear Sciences, University of the Witwatersrand,Private Bag 3, r-Vits 1050, Iohannesburq. South Africa1. MethodologyThe materials used in this work were PTW Diadosionisation chamber and electrometer, SIB mammographybreast substitute phantom, AGFA Mamorayscreen and film types, X-Rite densitometer and AGFAImaging Min/Med film processor with processingcycle and temperature of 90 sand 34.5 DC, respectively.For the incident air kenna method, mean glandulardose (MGD) values were determined from the tubeloading (TL) and the entrance surface air kenna (ESAK)measurements, following the procedure described bythe European Commission (CEC. 1996). Each electrometerreading was corrected for ambient temperatureand pressnre (Assiamah et al.. 2003a). The half valuelayer (HVL) for the selected TL was determined and theMGD calculated using the equationfrom the adjusted photon Iluence, ¢i(corrl from1.83 x 10- 11 ¢j{I:\}[r)Ei(flel"lt rX;(E)T PH = . (2), . PT,P,HWhere E, is photon energy of the lh energy bin; (!1el"l)1 ir isthe linear energy absorption coefficient of air at s.t.p.;PT,P,H is the density of air at temperature T, pressureP and humidity H (Cember, 1996). Absorption energycoefficient "spectra" of (1Ien)~lr were generated for eachenergy bin (Assiamah et al., 2003b). Using exposure toMGD conversion factor of 0.873, the MGD wascalculated (Wolbarst, 1993) from the exposure values byMGD ~ 0.873X,(E)T,P,H x g X p. (3)2. Results and discussionMGD = ESAK x g x p, (I) The ESAK and MGD values (Table 1) from bothtechniques showed slight differences (less than 5-12%)where 9 is the ESAK to MGD conversion factor at themeasured HVL and the p-factor converts ESAK for thePMMA breast substitute to that of the model breast(Dance et aI., (999). In calculating MGD from spectralenergy fluence, X-ray spectra measurements using thehigh-purity germanium detection system and correctiontechniques, as described in a companion paper (Mavundaet al., 2003), were used. Each of the spectra wasmodified to take into account the presence of the breastcompression plate and also the aluminium thicknessneeded to match the first HVL obtained there to thatfrom the ionisation chamber measurement for the sameTL Exposure, Xj(E)T,P,l! (in roentgen), was calculated"Corresponding author. Tel.: -;-27-11-717-7930; fax: +27­11-717-6932.E-mail address: assiamah':tIschonlan.src.wits.ac.za (1\1. As·siamah).for the thinner phantom thicknesses and lower kVpvalues. At 32kVp however, the ESAK and MGD fromthe spectral measurements differed by as much as 100%from the direct measurements, even though they havethe same ftrst HVL. The significant difference in theESAK and 1'v1GD values from the direct measurementand that from the spectral data was ascribed to thelarger component of scattered radiation incident onthe germanium detector. For the same phantom thickness,there was dose reduction in using higher kVpvalues than lower ones. The reduction was significantand dependent on phantom thicknesses. Fig. I shows thefilm contrast at 28 kVp with varying tube loading fordifferent PMMA phantom thicknesses for a range ofoptical densities measured using the procedure byDesponds et al. (1994). The contrast versus phantomthickness relationship was linear for thinner phantomsand exponential for larger ones.0969·806X i$ - see front matter D 2004 Elsevier Ltd. All rights reserved.doi: I0.10 16j.radphyschem.2004.05.00 I55

ELSEVIERAvailable online at www.sciencedireet.comSCIENCE@DIREcyeRadiation Physics and Chemistry 71 (2004) 957-958Radiation PhysicsandChemistrywww.elsevier.com/locate/radphyschemDosimetric techniques for mammography X-ray beamsM. Assiamah*, T.L. Nam, R.J. KeddyHealth Physics SeroicetSchonland Research Institute for Nuclear Sciences. University ofthe Witwatersrand,Private Bag 3, Wits 2050, Johannesburg, South Africa1. MethodologyThe materials used in this work were PTW Diadosionisation chamber and electrometer, SIB mammographybreast substitute phantom, AGFA Mamorayscreen and film types, X-Rite densitometer and AGFAImaging Min/Med film processor with processingcycle and temperature of 90s and 34.5°C, respectively.For the incident air kerma method, mean glandulardose (MGD) values were determined from the tubeloading (TL) and the entrance surface air kenna (ESAK)measurements, following the procedure described bythe European Commission (CEC, 1996). Each electrometerreading was corrected for ambient temperatureand pressure (Assiamah et aI., 2003a). The half valuelayer (HVL) for the selected TL was determined and theMGD calculated using the equationfrom the adjusted photon ftuence, cPi(corr) from1.83 x 1O-II¢i(COrrjEj(flen)1irXi(E)T P H = . (2), , PT,P,HWhere E;is photon energy of the /h energy bin; (Ilent r isthe linear energy absorption coefficient of air at s.t.p.;PT.P,fI is the density of air at temperature T, pressureP and humidity H (Cember, 1996). Absorption energycoefficient "spectra" of (,uen)1 IT were generated for eachenergy bin (Assiamah et al., 2003b). Using exposure toMGD conversion factor of 0.873, the MGD wascalculated (Wolbarst, 1993) from the exposure values byMGD ~ 0.873X,(E)T.P.il x g x p. (3)2. Results and discussionMGD~ ESAK x 9 xp, (I) The ESAK and MGD values (Table I) from bothtechniques showed slight differences (less than 5-12%)where 9 is the ESAK to MGD conversion factor at themeasured HVL and the p-factor converts ESAK for thePMMA breast substitute to that of the model breast(Dance et al., 1999). In calculating MGD from spectralenergy fluence, X-ray spectra measurements using thehigh-purity germanium detection system and correctiontechniques, as described in a companion paper (Mavundaet al., 2003), were used. Each of the spectra wasmodified to take into account the presence of the breastcompression plate and also the aluminium thicknessneeded to match the first HVL obtained there to thatfrom the ionisation chamber measurement for the sameTL Exposure, X;(Eh,p,H (in roentgen), was calculated"Corresponding author. TeL: +27-11-717-7930; fax: +27­11-717-6932.E-mail address:assiamahr.gschonJan_src.wits.ac.za(~1.Assiamah).for the thinner phantom thicknesses and lower kVpvalues. At 32 kYp however, the ESAK and MGD fromthe spectral measurements differed by as much as 100%from the direct measurements, even though they havethe same first HVL The significant difference in theESAK and MGD values from the direct measurementand that from the spectral data was ascribed to thelarger component of scattered radiation incident onthe germanium detector. For the same phantom thickness,there was dose reduction in using higher kVpvalues than lower ones. The reduction was significantand dependent on phantom thicknesses. Fig. I shows thefilm contrast at 28 kVp with varying tube loading fordifferent Pr.,,1MA phantom thicknesses for a range ofoptical densities measured using the procedure byDesponds et al. (1994). The contrast versus phantomthickness relationship was linear for thinner phantomsand exponential for larger ones.0969-806X/$~see front matter © 2004Elsevier Ltd. All rights reserved.doi: I0.1a16ij.radphyschem.2004.05.00I55

958 M Assiamah et at. I Radiation Physics and Chemistry 7l (2004) 957-958Table 1Comparison of ESAK and MGD calculated from measured air kenna values with that calculated from measured spectra at focal spotdetector distance of 65emNominal tube voltage (kV p) Phantom thickness (mm) Direct measurements Measured spectraESAK (mGy) MGD (mGy) ESAK (mGy) MGD (mGy)26 20 1.16 0.52 1.18 0.5326 3D 2.94 0.93 2.6 0.8128 2.51 0.85 2.64 0.8926 40 6.02 1.45 6.2 1.4828 5.23 1.35 5.5 1.353D 50 8.3 1.8 8.0 1.7232 6.4\ 1.44 12.9 2.832 60 13.41 2.62 25.8 5.0ea0.25• ~ 0,20.•,- ,TT/~to estimate MGD values. However, at higher Mo anodeX-ray tube voltages the spectral measurement method issignificantly more sensitive to the presence of scatterradiation. The best range of optical density was 0.9-2.0for thinner phantoms with equivalent contrast values of0.25-0.38 and for the thicker phantoms, a range of0.8-1.6 with equivalent contrast values of 0.2-0.34.U0.15ReferencescOC-5_2I1mmPMMA__ 30mmPMMA__ 40",,~ P"lMA_lC_ 50 m", PMMA__ 60m",PMMA01----,:,---,.::----,,,------,,:--::;;:------::1c'00 150 200Tube loading (mAs)Fig. l. Variation of contrast with TL for different thicknessesof PMMA phantom at a tube voltage of 28 kVp'The contrast of the 60 mm phantom was low andassumed to be due to reciprocity law failure. (NB togenerate a stable, latent image sensitivity speck a grainof silver halide must accumulate a "critical mass" ofsilver atoms. If exposure rate is too low, i.e. longexposure times and therefore low flux, some silver atomscould be freed and leak away from the sensitivity speckresulting in criticality not achieved. This behaviour isreferred to as reciprocity law failure. In the case ofmammography for a given tube voltage and current,films become lighter as a result of long exposure times(Wolbarst, 1993).)The study has shown that the incident air kermamethod and spectral energy fluence method can be usedAssiamah. M., Mavunda, D., Nam, T.L., Keddy, R.J., 2oo3a.Effect of pressure, temperature and humidity in air onphoton fluence and air kerma values at 10\\' photon energies.Radial. Phys. Chern. 68 (5), 707-720.Assiarnah. M., Mavunda. D., Nam, T.L., Keddy, R.J., 2003b.Segmented multifit of polynomial function for massattenuation and energy-absorption coefficient values.Radial. Phys. Chern. 67, 1--6.CEC (Commission of the European Communities), 1996.European protocol on dosimetry in mammography. EUR16263, European Commission, Luxembourg.Cember. H.. 1996. Introduction to Health Physics, 3rd Edition.l\1cGraw-Hill, New York, pp. 169--179.Dance, D.R., Skinner, C.L., AIm, C.G., 1999. Breast dosimetry.App!. Radial. Isot. 50, 185-203.Desponds. L., Klausz, R., Heidsieck, R., 1994. Imagequality and glandular dose for rhodium K-edge filterin mammography: performance comparison of therhodium anode with molybdenum anode. Coolidge Sci.Rev. l. 7-12.Mavunda, R_D., Assiamah, M., Nam, T.L., Keddy, R.J., 2003.Bremsstrahlung spectra from diagnostic X-rays. A paperpresented at ISRP-9. Cape Town, RSA, October 2003.Wolbarst, A.B., 1993. Physics of Radiology, Vol. 185,Int. Edition. Appleton and Lange, Connecticut, USA,pp. 121-131.56

Chapter 3SEGMENTED MULTIFIT OF POLYNOMIALFUNCTION FOR MASS ATTENUATION ANDENERGY-ABSORPTION COEFFICIENT VALUESRadiation Physics and Chemistry 67 (l), 1-6.57

PERGAMONAvailable online at www.sciencedirect.comSCIENCE@DIRECTeRadiation Physics and Chemistry 67 (2003) 1-6Radiation PhysicsandChemistrywww.elsevier.comllocate/radphyschemSegmented multifit of polynomial function for massattenuation and energy-absorption coefficient valuesM. Assiamah, D. Mavunda, T.L. Nam, R.J. Keddy*Health Physics ServicelSchonland Research Institute for Nuclear Sciences, Unicersity of the Witwatersrand, Priuue Bag 3, Wits 2050,Johannesburg, South AfricaReceived 20 December 2002; accepted 28 December 2002AbstractThe photon energy region 1-20keY cannot be ignored for any theoretical beam-path attenuation and/or dosimetrycalculations for low energy (mammography, dental units, etc.) bremsstrahlung spectra. The polynomial functionpresented by Tucker et a]. for fitting the mass attenuation coefficient data for dry air (assumed at temperature of 20 c eand pressure of 1.013 x 10 5 pa) has therefore been adapted to the photon energy region below 20keV. This region wasexcluded by Tucker et at. It is found that only by dividing the 1-20keV region up into smaller sub-regions to obtainvalues for the polynomial parameters could a good fit be obtained over the whole of the region from I to 200keV.1J 2003 Elsevier Science Ltd. All rights reserved.Keywords: Bremsstrahlung; Attenuation coefficients; XvraysI. IntroductionAccurate values of mass and mass-energy attenuationcoefficients are essential when quantifying the interactionsof an X-ray beam with matter. X-ray spectrameasurements, exposure or air kerma and dose calculations,are areas of both diagnostic and therapeuticradiology where these attenuation coefficients areemployed. It is well known that the coefficients arefunctions of both the photon energy and the physicalproperties of the irradiated material. Hubbell (1969,(982), Cullen et aI. (1989), Seltzer (1993), Seltzer andHubbell ([995), and Hubbell and Seltzer (1996) andmany other authors have compiled mass attenuationdata. These compilations have been used in bremsstrahlungspectra model. exposure and dose calculations. Themass attenuation compilations do not however coverevery energy value in the energy ranges normallypresented. To interpolate values for the mass attenua-"Corresponding author. Tel.: + 27-11-717-6923; fax: -4- 27­11-7l7-6932.E-mail address:keddy@src.\\..its.ac.z-a (R.J. Keddy).tion coefficients for energies of interest, other than thosetabulated, the compiled data are often fitted utilisingleast-square techniques and an appropriate mathematicalexpression. A mathematical expression given byTucker et al. (1991) for modelling tungsten targetspectra is recommended as being suitable for fittingmass attenuation data. The Tucker et al. expression isbed ep(x) ~ Q + x'6 + x2.7 + xl.5 + xl 5where x = £/lOO, E ~ X-ray photon energy (keY) and Q.b. c, d, and e are derived parameters. Of particularinterest for exposure and dose calculations for 10\~,:energy spectra e.g. mammographic exposures, is theregion below 20 keY; a region not addressed by Tuckeret al. They address the region from 20 to 200 kev. Forthis lower end of bremsstrahlung energies, air attenuationcannot be ignored for theoretical computations ofabsorbed dose and/or exposure values. The Tucker et al,expression has been used therefore to fit mass attenuationcoefficient data as well as mass energy-absorptioncoefficient data, for dry air in the region below 20 keYusing a least-square technique. The mass energyabsorptionand mass attenuation coefficients data used(I)0969-806X/03/S-see front matter 1'j 2003 Elsevier Science Ltd. All rights reserved,dodO.1016/S0969-806X(03jOOOII-258

2 M Assiamah et al. I Radiation Physics and Chemistry 67 (2003) 1-6for this exercise were those compiled by Hubbell andSeltzer (1996) now very conveniently assembled byGerward in a Windows platform programme calledWinXCom.2. MethodologyFor the total region looked at in this work, datapoints from 1 to 200keV were .used, Three types ofapproximation approaches were adopted for fitting thedata.(I) Fitting the entire suite of data values once for oneset of polynomial parameter values.(2) Fitting the data values in two regions. From 1 to20keV and from 20 to 200 keY to obtain two sets ofparameter values, one for each region.(3) Dividing the 1-200keV regions into four separateregions and fitting the data points in each subregionseparately for four sets of parameter values.The mass energy-absorption and mass attenuationdata were fitted together with the K edge. (It isimportant that data points below and above the Kedge are fitted separately and that the values of theedges be used as boundaries ofcontiguous regions.]The derived parameters from our fitting proceduresas well as those given by Tucker et al. for massenergy-absorption and mass attenuation coefficientsfor air were used to compute attenuationTable IDerived constants a, b. c, d, and e including Tucker et al. in cm 2/g for air mass energy-absorption coefficientsEnergy range (keY) a b c d eSingle fit1-200 11.599 -3.38£-01 2.24£-02 -2.34£-04 8.22E~07Double fit1-20 -1.7431 0.064174 6.97£-03 3.33£-04 -1.58£-0620--200 3.10£-02 -1.88£-02 1.29E-02 -1.93£-03 2.l2E-04MUltiple fit1-3.203 343.78 -4.3313 1.26£-01 -3.33£-03 1.18£-053.203-8 -9.7059 6.92£-01 -6.02E-02 5.94£-03 -7.14£-055-20 1.03£-01 -2.66£-02 8.38E-03 5.25E-04 -5.IIE-0620--200 3.10£-02 -1.88E-02 1.29E-02 -1.93E-03 2.12E-04Tucker et al. constants20--200 2.892E-02 -1.I70E-02 5.836E-03 7.588E-04 -7.892£-06Table 2Derived constants a. b. c. d, and e including Tucker et al. in cm 2ig for air mass attenuation coefficientsEnergy range (keY) a b c deSingle fit1-200 -2.35E-01 4.17E-02 7.96E-03 2.96£-04Double fit1-20 -1.9125 8.66£-02 6.30£-03 3.56£-0420--200 1.04£-01 8.34E-02 -5.28£-02 2.0IE-02Multiple fit1-3.203 3.7989 -4.761 1.37E-01 -3.67£-033.203-8 -11.13 8.10£-01 -7.07E-02 6.76£-035-20 2.24E-01 -1.13£-02 6.67£-03 7.17£-0420--200 1.04£-01 8.34£-02 ~5.28E-02 2_01 E-02Tucker et al. constants20--200 1.088£-01 6.004£-02 ~2.5S1E-02 8.473£-03-1.67£-06-IAO£-031.30£-05-8.10£-058.36£-06-1.40£-03-3613£-0459

M Assiamah et at. I Radiation Physics and Chemistry 67 (2003) /-6 3coefficient values. The results from the computationhave been compared to the original massenergy-absorption and mass attenuation coefficientsdata compiled by Hubbell and Seltzer (1996).3. Results and discussionThe parameters a, b, c, d, and e derived from ourfitting exercise and those presented by Tucker et al. forair, in units of cm 2/g, are presented in Tables 1 and 2 forthe mass energy-absorption coefficients and massattenuation coefficients data, respectively. Figs. 1--4illustrate the fits graphically, for the regions 1-20 and20-200 keY which also includes the results of the fitusing the Tucker et al. parameters.The four sub-region energy range fit of the massenergy-absorption coefficients data gave the lowerdeviations and are in much better agreement with theoriginal data than the single or two energy range fits.Fitting data points of 1-20keV as one-region, results ineoooI~HUB8E.ll DATA I---MULTIPLE FIT.'" Ee. aooc~'a~ 25008• o;; 2000~o~~ 1500~•:: 1000~\-,oo z , zoEnergy (keV)Fig. 1. Comparison of mass energy-absorption coefficient data by Hubbell and Seltzer with values obtained from multiple energy fit,for the energy range 1-20keV.'"00350 0I=---HUBBELL DATA I---"IIULT'PLE FIT0"~ 300"~2500" ~8• '00o" , c 150~00·~ 1000500" -;,02610 12ts te .Energy (keV)Fig. 2. Comparison of mass attenuation data by Hubbell and Seltzer with values obtained from multiple energy fit, for the energyrange 1-20keY.60

4.,.,M. Assiamah et af. I Radiation Physics and Chemistry 67 (2003) 1-6I~BBEUOATA_TUCKER eI!aI CONSTANTS"'ULTIPLE FITf i 0.4u7i8 o.r: i c.,\"',,--- .o au 100 120Energy (keV}'" ". '".Fig. 3. Comparison of mass energy-absorption coefficient data by Hubbell and Seltzer with values obtained from multiple energy fitand Tucker et al. constants, for the energy range 2o-200keV.09o.e1=-If.JB6ELL DATA___TIJCKERet aI CO/

M. Assiamah et al. I Radiation Physics and Chemistry 67 (2003) /-6 580I 60,:c• 40,g:;~ 20~"E 0•Iie,-20//I /1. ----/

6 M Assiamah et al. I Radiation Physics and Chemistry 67 (2003) 1-6eo00 "'~'n~'F "0' .~'>0.;' /.e.1_~-,»>-----I__MULTIPl.£FI~--SINGLE FIToooece FITV '\ /0.0Energy {keV}~ /Fig. 7. Deviation of the mass attenuation coefficient data derived from the single, double and multiple energy fit, from the originalvalues by Hubbell and Seltzer for the energy range 1-20keV.ta0/-z o \r=--TUCl(ER eI aI CONSTANT~sr-.I ~ 77s.--. I I v-, Io/ \\ //s\ ~ //c\ 7~ut.T1PLEFIT.0I,.s,"'"

Chapter 4EFFECT OF PRESSURE, TEMPERATURE ANDHUMIDITY IN AIR ON PHOTON FLUENCE ANDAIR KERMA VALUES AT LOW PHOTONENERGIESRadiation Physics and Chemistry 68 (5), 707-720.64

PERGAMONAvailable online at www.sciencedireet.comSCIENCE@DIREcyaRadiation Physics and Chemistry 68 (2003) 707-720Radiation PhysicsandChemistrywww.elsevier.comllocate/radphyschemEffect of pressure, temperature and humidity in air on photonfluence and air kerma values at low photon energiesM. Assiamah, D. Mavunda, T.L. Nam, R.J. Keddy*Health Physics Service, Schonland Research Institute for Nuclear Sciences, University ofthe Witwatersrand, Private Bag 3, Wits 2050,Johannesburg, South AfricaReceived 3 April 2003; accepted 19 May 2003AbstractAn investigation into the effect of pressure, temperature and humidity in air on photon fluence at a typicalmammography, low bremsstrahlung energy (25kVp), has been carried out. Pressure values corrected for humidity atvarying temperatures were employed to determine the density of moist air. Using the corrected moist air densities, theX-ray photon fluence remaining after air attenuation has been computed assuming an X-ray focal spot-detectordistance of 65 CID. Bremsstrahlung spectral distributions and the effects of pressure, temperature and humidity on thephoton fluence from molybdenum and tungsten targets are illustrated. Comparative results suggest that such effectscould be significant and need to be considered when calculating exposure levels from low-energy photons. Theinvestigation showed that air kerma values from an X-ray spectrum that has significant lower-energy components islikely to be more sensitive to changes in pressure, temperature and humidity than the air kerma from an X-ray spectrumwith lower-energy components less pronounced. Non-negligible air kerma values are involved.© 2003 Elsevier Ltd. All rights reserved.Keywords: X-ray photon fiuence; Attenuation; Mammography; Bremsstrahlung; Air kenna1. IntroductionMammography X-ray examinations are typicallyperformed at a focal-spot detector distance (FSD) of55--65em. In the calculation of the photon fluencereaching the detector for dose or exposure measurementsat mammography energies, the attenuation due toair for the specific FSD must be taken into consideration.In many such calculations, the air is normalIyassumed to be dry; hence it is common to find thedensity of dry air at sea level being applied in suchcomputations. However, sea level atmospheric pressuresdo not apply with many measurements being conductedat centres at significant altitudes. Air normally containswater vapour and is classified as dry or humid based on"Corresponding author. Tel.: +27-11-717-6923; fax: +27­1I-717-6932.E-mail address:keddy@src.wits.ac.za (R.J. Keddy).its water vapour content. The humidity is specified usingpartial pressures of water in the air. The relativehumidity of the water content in the air is not constant,varying from a minimum of zero to a maximum of100%, determined by the saturated vapour pressure ofwater at the given air temperature. The density of air istherefore affected by changes in environmental conditions,i.e. temperature, pressure and moisture orhumidity. This study sets out to determine the effect ofthese physical properties on the photon fluence and airkerma values of mammography bremsstrahlung beams.2. MethodologyThe photon fluence (photons-ern") used for this studywas obtained from a computer generated X-ray spectrumcode namely: molybdenum anode spectramodel using interpolation polynomial (MASMIP) and0969-806X/03jS-see front matter~' 2003 Elsevier Ltd. All rights reserved.doi; 1O.!0 !6jS0969-806X(03)00398-O65

708 M Assiamah er aL I Radiation Physics and Chemistry 68 (2003) 707-720tungsten anode spectra model using interpolationpolynomial (TASMIP) developed by Boone et a!.(1997). The X-ray spectrum for a 25 kYp tube potentialwas generated from both the MASl\lIP and TASMIPmodels. A nominal filtration of 0.03 mm molybdenumwas applied to the molybdenum target spectrum, whilsta 0.05 mm rhodium filtration was applied to the tungstentarget spectrum. The spectral shaping (filters) materialsapplied are routinely used on clinical mammographysystems. A focal spot-detector distance of 65 em (anFSD commonly used at clinical mammography settings)was assumed for the calculation. For computationalpurposes the bremsstrahlung spectrum was divided into0.5 keY energy bins. The photon fluence remaining ineach bin after air attenuation was calculated byapplying, to each bin, the energy-dependent Lambert­Beer's law equationwhere if> is the photon fluence (photon/em") in each binafter attenuation by air of thickness e (g/cnl) with massattenuation coefficients It (cm 2jg) and ¢o is the initialphoton fluence before attenuation (Archer and Wagner,1982; Attix, 1986; Wolbarst, 1993). The thickness, isgiven by the product of the density p in g/cm? and thefocal spot-detector distance x in em. The density of airused in Eq. (I) has been corrected for temperature,pressure and humidity. The corrected air densities usedfor this investigation are those given in the CRCHandbook of Chemistry and Physics (l983~1984) for atemperature range of 5-3SOC and corrected pressurerange of 600--780mrnHg. The densities for moist air attemperatures 40°C, 45°C and 5D oC, not given in thereference for the specified corrected pressures, werecalculated using Eq. (2) and Table II of section F ofWeast et a!. (l983~1984).The density PT,P,H of moist air at absolute temperatureT. barometric pressure P in mmHg, is given by= I 2929 (273.13) (P - 0.3783e)Pr,p.H· T 760'where 1.2929 is the density of dry air in gil at standardtemperature and pressure (s.t.p.) i.e. 0'Cf273.13 K and760 mmHgjlDI.3 kPa, respectively. The parameter "e'appearing in Eq. (2) is the vapour pressure of themoisture in the air in millimetres. Employing the dewpointmethod and the vapour pressure of water atdifferent dew points, the humidity H of the air has beenaccounted for (Marion and Hornyak, 1982). Thecorrected pressure values at various temperatures werethen employed to determine the density of moist air forthe calculated relative humidity. Relative humidities of0--100% at intervals of 10% were used in the calculations.Using the photon fluenee if>i{COIT)' that which remainsafter 65em air attenuation, exposures X(E)T,P,H in(I)(2)roentgen (R) at absolute -temperature T, barometricpressure P and humidity H was calculated employing_1_.8_3_x_ l_0_-_ 11 ...: "'"

M Assiamah et al: I Radiation Physics and Chemistry 68 (2003) 707-720 709300.....-0--600 mnli;l-620mnli;l250-+-640mnli;l~x-660mnli;l200-+-680mnPg--a-700mnli;l-nOmnHg!SO-o-740mn!i;JX--780mnti'j:10050a~+~~+-+ :- .-'-+-.~.-.~.'I--'·)!,"'~~~~~~~'~a 1.00.0

710M. Assiamah et al. I Radiation Physics and Chemistry 68 (2003) 707-72050~~ .,..0~cma:I:30E0~~P.a-0-10 ~ 20• ~uc=_600mnHg-620mnHg......-6~OmnHg-X-660 ImlHg't\-------------------------------i--680ImlHgX-700rrmHg-+-nOrrmHg-740mnHg-6--780 mmHg• ~=c~s:~ •E 001=~~g 0.m~0,-101215IB•en",gy (keV)2127Fig. 3. Deviation of the photon fluence due to pressure variation from the photon fluence calculated at aoc, 760mmHg and 0%relative humidity for a tungsten target at 25 kVp.2Dr-----------------------------------,~ 10e~~aos-

M. Assiamah e! aJ. J Radiation Physics and Chemistry 68 (2003) 707-720711180 r---------------------[~;;____==;;:;:l __5"C_1l aC160}----------------------------1--+-I5"C-Z-25"C-£f-2QOC_30°C140Jl------------------------------I--6-35"C--+-4QoC~~--45"C-'"_~oC: § 120Jll-------------------------------'=======r-lo ~~ ~~~ o ~]i~0. 0E'"e tij-:x:- '"~Eg~.~ ~

~~712M Assiamab et al. I Radiation Physics and Chemistry 68 (2003) 707-720as300"-e c•~:r25E0~~· o~ ~c."•.2:m sp~ 2Do~iiiEe- e.l\- --i-t3-ZSOC -30·C- ..soc _5JOC~E D0'=~cg S-" >•a,fiEEnergy {keV}212730Fig. 7. Deviation of the photon fluence due to temperature variation from the photon fluence calculated at aoc, 760mmHg and 0%relative humidity for a tungsten target spectrum at 25 kVp.is14-:~~g-'"~aIIWDBD~-:.r:-:/0.4-:02DD5 10 15 zo 25 30Temperature (0C)35 40 45 50 55Fig 8. Overall deviation (%) due to temperature on the integrated photon fluence from the photon fluence calculated at DOC,760mmHg and 0% relative humidity for a tungsten target spectrum at 25 kVp.70

aorEi!lM. Assiamah et al. I Radiation Physics and Chemistry 68 (2003) 707-720 713Llll,--------------------f-:--=--=---::::;,--:ol\I -~__~ -0-40%--a-SO%.-o-6OOfo-+-- --",",f------------------------------l ~....100lUll~•o"+"s.1!];i0ạ1!Eg~~~0(l8J -----.-MO020•14 J6Energy{keV)Fig. 9. Deviation of photon fluence due to a humidity variation from the photon fluence calculated at Qce and 760mmHg for amolybdenum target at 25kVp.O.lJ3S00'"0.Q25z 0.Q20cg.!!1~OO~ C0.010OJlOS.>.::-:-:.: .:OJlOOo 10'" '"40 so 6Q 70 .0 90 100 110I-imdity (%)Fig. 10. Overall deviation (%) due to humidity on the integrated photon fluence from the photon fluence calculated at Dce and760mmHg for a molybdenum target at 25kVp.71

714 M. Assiamah et aL I Radiation Physics (DId Chemistry 68 (l003) 707-72005D% 20%--]0% --t3-40%0.4-e-SO% -60%~ 0.4~Ie ~ 03cmp o1ii0.3D! 02~~ 0.2~cg 0.1.!l!Ii;oOD+ .'*'*~\.~.\'~\~.~~~• ~+:+:-::~r-......2~~2: s - - -61215•--~~~ -- ---18&1ergy (keV)• I$J\lrl'l9l92124~70% --80%-90% _1l0~Fig. II. Deviation of photon fluence due to a humidity variation from the photon fluence calculated at Goe and 760 mmHg for atungsten target at 25kVp.'"25,.--------------------------------,30OD20f-----------------------------------j§:.'15cg.a >• a'"10/-:OJlOO -I--~--_-_--_--_-_--_-_--_-_-o 10 20 30 40 50 60HlnIidItyN_ _I70 80 50 100 noFig. 12. Overall deviation (%.) due to humidity on the integrated photon fluence from the photon fluence calculated at Qce, 760 rrunHgat 0% relative humidity for a tungsten target at 25kVp.72

M Assiamah et a/. I Radiation Physics and Chemistry 68 (2003) 707-720 71514E+09UE+09I-A-·IlDE+09ị!10cB 8.oE+080s:.!!o euc~.~ 6.DE+08~c015s:11.4.QE+082.OE+OBO.DE+08~fl/LJ.>4 6 10 12 14 16 III 20 22 24 26•Energy (keV)Fig. 13. A comparison of the photon fluence spectrum from a molybdenum target at 25kVp. (A) Spectrum at 760mmHg, O°Cand 0%relative humidity. (B) Spectrum corrected for temperature (3Y'C), pressure (600mmHg) and relative humidity (50%).0.450.40I-A-·I035030>: lJ.25o.em~ 010~OJOODSoro~-:~ 1\r>:,"Jf"1~ 015-llDS4 6 8 10 12 14 1£Energy (keV)18 20 22 24 26Fig. 14. A comparison of the air kerma spectrum from a molybdenum target at 25kVp. (A) Spectrum at 760mmHg, OCC and 0%relative humidity. (B) Spectrum corrected for temperature (3S 0C), pressure (600mmHg) and relative humidity (50%).73

716 M Assiamah et at. I Radiation Physics and Chemistry 68 (2003) 707-72025E+(lS 1------:--------------------------;::======:::1I-A --IZ.DE+08 f--------------------~"-----~._---------jL5E+

At. Assiamah et al. I Radiation Physics and Chemistry 68 (2003) 707-720 717Table 2Comparison of contributions of three arbitrary energy segments to the overall air kenna in cGy from a molybdenum spectrum at25kVp, DoC,760mmHg and 0% relative humidity with the molybdenum target spectrum at the same tube voltage and corrected fortemperature, pressure and humidityContribution to airkenna by energyrange 4-8 keVContribution to airkenna by energyrange 8-l5keVContribution to airkenna by energyrange 15-25keVAir kenna (cGy) at760mmHg,DOC,O% RHTotal contributionPercent contribution (%)0.0060.310.91250.430.98554.46Air kenna (cGy) at600mmHg35°C, 40% RHTotal contributionPercent contribution (%)% difference in airkerma due to P, T, Hcorrection0.0110.3887.521.41651.5555.231.46353.2548.48The overall percentage due to the combined effects oftemperature, pressure and humidity is also shown. Overall difference in air kermadue to P, T and H correction =51.85%.Table 3Comparison of contributions of three arbitrary energy segments to the overall photon fluence from tungsten spectrum at 25 kVp, O°C,760 mmHg and 0% relative humidity with the tungsten target spectrum at the same tube voltage and corrected for temperature,pressure and humidityContribution to photonfluence by energy range6-8keVContribution to photonfluence by energy range8-1 5keVContribution to photonfiuence by energy range15-25keVPhoton fluence at760mmHg,DOC, 0% RHTotal contributionPercent contribution (%)1.15E+040.00034.75E+0813.973.04E+0989.32Photon fluence at600mmHg35"C, 40% RHTotal contributionPercent contribution (%)% difference inphoton ffuence due toP, T, H correction1.47E+040.000427.485.0IE+0814.355.463.IIE+0988.992.29The overall percentage due to the combined effects of temperature, pressure and humidity is also shown. Overall difference in estimatedphoton fluence due to P, T and H correction = 2.68%.were selected: 4-8, 8-15 and 15-25keV for the molybdennmspectrum and 6-8, 8-15 and 15-25keV for thetungsten spectrum. The overall difference in theestimated air kerma values of the entire molybdenumand tungsten spectra due to the combined effects oftemperature, pressure and humidity correction is alsoshown in Tables 2 and 4.3.1. PressureAs expected, the effect of pressure was more pronouncedat the lower energies than at higher energies (Fig. I).For the molybdenum spectrum, a deviation range fromless than 1(}.-260% was observed, between the energies 4and 10 keY, for a pressure range of60G-700 mmHg. Forthe same energy range, the deviation was 2-50% forpressure values 72G--780mmHg. Above lOkeY, thedeviations due to pressure effects ranged from less thanI% to 10% for the entire pressure range of 600­780 mmHg. The overall effect of pressure on theintegrated photon fiuence is minimal with a maximumof 3% at 600 mmHg pressure and a minimum of 0.33%at 780mmHg (Fig. 2). The low overall deviation ofpressure on the photon fIuence in spite of the highdeviations at the lower energies is due to the fact thatafter filtration of the spectrum, the lower-energy75

718 M. Assiamah et all Radiation Physics and Chemistry 68 (2003) 707-720Table4Comparison of contributions of three arbitrary energy segments to the overall air kenna from a tungsten spectrum at 25 kVp, aoe,160mmHg and 0'%, relative humidity with the tungsten target spectrum at the same tube voltage and corrected for temperature,pressure and humidityContribution to airkenna (cGy) by energyrange &-.8 ke VContribution to airkenna (cGy) by energyrange 8-15keVContribution to airkerma (cGy) by energyrange 15--25 keYAir kenna (cGy) at760mmHg,o-c, 0% RHTotal contributionPercent contribution (%)1.32E-050.00180.1824.960.5779.74Air kenna (cGy) at600mmHg:35"C, 40% RHTotal contributionPercent contribution (%)% difference in airkenna due to P, T, Hcorrection2.42E~050.002384.130.2725.6052.940.8479.1347.96The overall percentage due to the combined effects oftemperature, pressure and humidity is also shown. Overall difference in estimatedair kenna (cGy) due to P, T and H correction =49.2%.component is removed hence its contribution towardsthe entire spectrum is reduced. The deviation pattern issimilar for both molybdenum and tungsten targets (Figs.3 and 4). The figures involved are however different. Theeffects on the tungsten target spectrum being lower thanthat from the molybdenum. The difference in thepressure effect on the molybdenum and tungsten targetsis as a result of the molybdenum X-ray spectrum havingmany more lower-energy components when comparedto the X-ray spectrum from tungsten.3.2. TemperatureThe temperature effect on the photon fluence increaseswith increasing temperature. Below 10keY, andfor temperature range of 5-25°C, a deviation of 1-70%was calculated for the molybdenum target (Fig. 5). Forthe same energies, the deviation was about 5~120% attemperatures 3Q-40°C. At temperatures above 40°C,deviations of 5% to more than 150% were observed forthe same energy range. At energies greater than lOkeY,the deviations were much lower and ranged from lessthan 1-10% for the entire temperature range, 5-50 c C.The overall temperature effect on the photon fluencefrom a molybdenum target yielded a deviation of 0.25­2% (Fig. 6). The patterns of the temperature effect onboth the molybdenum and tungsten targets were similar(Figs. 7 and 8). The temperature effect was howeverlower for the tungsten target than for the molybdenumtarget. This may be attributed to the proportionallylarger contribution from the low-energy componentsthat are present in the molybdenum spectrum whencompared to tungsten X-ray spectrum.3.3. HumidityThe humidity effects on the photon fluence for themolybdenum and tungsten targets were very lowcompared to the pressure and temperature effects. Thedeviations due to humidity effects on the entire energyspectrum of both the molybdenum and tungsten targetswere less than 5% (Figs. 9 and II) for the full relativehumidity range, 0-100%, at all temperatures considered.3.4. Pressure. temperature and humidityThe photon fluence comparison (Table I) showed thatthe effect of the pressure, temperature and humiditycorrection on the three arbitrary energy ranges chosenfor the molybdenum target varied with the effect beinghighest (29.5%) in the lower energy range; 6.6% and2.7% for the energy ranges 8-15 and 15-25keV,respectively. The overall difference in photon fluenceof the molybdenum spectrum due to the correction forP, T and H was 4%. The overall difference was low inview of the fact that the contribution to the lower-energysegment of the spectrum from molybdenum is reduceddue to filtration. The photon fluence comparison for thetungsten spectrum (Table 3) followed a similar patternto the molybdenum spectrum with the overall differenceeven smaller.The results from the air kenna calculations of themolybdenum spectra (Table 2) indicate that at theenergy range 4-8 keY, the combined effects of pressure,temperature and humidity is significant. The contributionof air kerma for this energy range to the air kermaof the entire spectrum is however negligible. Thecontribution to the calculated air kerrna in the energyrange 8-15keV was 52% for the molybdenum spectrum76

At. Assiamah et al. I Radiation Physics and Chemistry 68 (2003) 707-720 719corrected for air density and 50% for the referencemolybdenum spectrum. For the same energy range, thecontribution to the calculated air kerma was 26% and25% respectively. for the corrected and referencetungsten spectra (Table 4). The overall percentagedifference in the estimated air kerma value due to thepressure, temperature and humidity correction at thisenergy range was 55% and 53%, respectively, for themolybdenum and tuugsten spectra (Tables 2 aud 4). Forthe energy range 15-25keV, the contribution to theestimated air kerma was 53% and 54%.,respectively, forthe corrected and reference molybdenum spectra; 79%and 80%, respectively, for the corrected and referencetungsten spectra. The overall difference in the estimatedair kerma at this energy range, due to pressure,temperature and humidity correction was 48% for boththe molybdenum and tungsten spectra. The overalldifference in the estimated air kerrna of the entirespectrum due to the combined effects of pressure,temperature and humidity was 52% and 49%, respectively,for the molybdenum and tungsten spectra at DOC,760 mmHg and 0% relative humidity shown in Figs. 14and 16 (Tables 2 aud 4), respectively. For the molybdenumspectrum, the air kerma difference contrastssignificantly with the 45% overall difference obtainedroutinely by correcting for pressure and temperatureonly (Table 5).4. ConclusionThe investigation has shown that the effect ofpressureon photon fluence of an X-ray spectrum from bothmolybdenum and tungsten targets are significant at7.QE-02I-A ~'I6.OE-025.QE-024.oE-02>:o 2- 3.QE-02m~ x~ 2.llE-021.OE·02O.lE+OO-l.(E-026 8 10 12 14 16 18Energy (keV)20 22 24Fig. 16. A comparison of the air kenna spectrum from a tungsten target at 15 kVp. (A) Spectrum at 760mmHg, O"C and 0% relativehumidity. (B) Spectrum corrected for temperature (3Y'C), pressure (600mmHg) and 50~/O relative humidity.Table 5Comparison of change in the air kenna values obtained using normal temperature and pressure correction with values obtained fromthe correction procedure outlined in this studyX-ray target materialPredicted air kenna change (%)Expected air kenna changeafter normal P, T correction (%)Predicted-Expected (%)MoW524945457477

720 M. Assiamah et al. I Radiation Physics and Chemistry 68 (2003) 707-720photon energies less than 10 keY. Lower-pressure valuesshowed higher deviations on the photon fluence at thisenergy range than higher-pressure values. Notwithstandingthis observation, the contribution of the lowerenergysegment to the entire spectrum is reduced afterfiltration; hence, the overall deviation due to pressure ofthe integrated photon fluence was found to be small. Ithas also been established from this study that thetemperature effect on photon ftuence for both targetmaterials are significant at the lower-energy segment ofthe X-ray spectrum. The temperature effect on thetungsten spectrum is however less. than for the molybdenumtarget. The moisture content of the air has beenfound not to have any appreciable effect on photonfluence at any energy segment of the spectrum unliketemperature and pressure. The study has establishedthat for high-altitude areas where air pressures are low,there could be a 4% difference in photon fluence whentemperatures ate high. It has also been established thatfor the same atmospheric conditions (low pressures andhigh temperatures) there could be a 50% overalldifference in estimated air kenna values in mammographyor low bremsstrahlung energy beam measurementsunless the air density is corrected for pressure andtemperature.The results presented suggest that unless air density iscorrected for atmospheric conditions at the time ofmeasurements, the overall difference in the estimated airkenna could be large. In the case of an X-ray spectrumwith a high proportion of low-energy contributions,unless the effect of temperature, pressure and humidityon the photon flux is considered over the entire energyrange, errors in kenna values significantly larger thanvalues obtained after normal routine temperature andpressure corrections could result. This can partly beattributed, to changes in the photon fluence spectrumresulting from changes in atmospheric conditions andpartly due to the influence of the energy and the linearenergy-absorption product term (E;)(J1enP), shown inEq. (3).AcknowledgementsThe authors wish to express their gratitude to Dr. J.M. Boone, Dr. T. R. Fewell and Dr. R. J. Jennings whomade available their molybdenum and tungsten anodespectral models used for this study.ReferencesArcher, B.R., Wagner, L.K., 1982. Med. Phys. 9, 844-847.Assiamah, M., Mavunda, D., Nam, T.L., Keddy, R.J., 2003.Radial. Phys. Chern. 67, 1--6.Attix, F.H., 1986. Introduction of Radiological Physics andRadiation Dosimetry. Wiley-Interscience, New York, pp.39-46.Boone, J.M., Fewell, T.R., Jennings, R.l., 1997. Med. Phys. 24,1863-1874.Cember, H., 1996. Introduction to Health Physics (3rdEdition), McGraw-Hill, New York, pp. 169-179.Gerward, L, Guilbert, N., Jensen, K.B., Levring, H., 2001.Rad. Phys. Chern. 60, 23-24.Hubbell, 1.H., Seltzer, S.M., 1996. Tables of X-ray MassAttenuation Coefficients and Mass Energy-AbsorptionCoefficients from I keY to 20MeV for Elements Z= I to92 and 48 Additional Substances of Dosimetric Interest.National Institute of Standards and Technology, USDepartment of Commerce, Gaithersburg, MD 20899.Marion, J.B., Hornyak, W.F., 1982. Physics for Science andEngineering, Part 1. Saunders College Publishing, NewYork, pp. 655-{;58.Weast, R.C., Astle, M.J., Beyer, W.H. (Eds.), 1983-1984. CRCHandbook of Chemistry and Physics, eRC Press, BocaRaton, FL, pp. F-9, F-IO.Wolbarst, A.B.,_ 1993. Physics of Radiology, InternationalEdition. Appleton and Lange, Connecticut. USA, pp. 96,122-131.78

Chapter 5SYNTHETIC DIAMOND AS A RADIATIONSENSING ELEMENT FOR MAMMOGRAPHYX-RAY BEAM DOSE MEASUREMENTS79

AbstractThe desirable properties of diamond have made the mineral a material of choice in radiationmeasurements and are currently used extensively in high-energy physics. Its use in the lowenergy beam such as mammography X-ray beam however, has not been fully investigated. Inthis presentation a diamond probe has been constructed using both chemical vapour deposited(CVD) synthetic diamond-as well as diamond produced under high-pressure high-temperature(HPHT) as the radiation sensing materials. An investigation into the exposure geometry of thediamond sensor that will give maximum absorption of the incident X-ray beam was alsoconducted in this study by calculations. The diamond samples were characterized to obtaininformation about the level ofimpurities especially nitrogen levels and consequently establishthe material quality. Nitrogen quantities in the diamond lattice have been shown to have aprofound effect on the. radiation detection properties of diamond (Nam et al., 1991). Thediamond sample surfaces were metallized with titanium, platinum and gold to provide ohmiccontacts. The probe was connected to both the Wellhofer Dosimetrie (model CD 500) and thePTW Unidos E commercial electrometers. In all the measurements, the incident radiationbeam was perpendicular to the edge of the CVD diamond plate to optimize absorption of theX-ray beam by the sensing material. The probe was calibrated against PTW Diados secondarystandard dosimeter. The results ofthe study are presented in both tabular and graphical forms.5.1. IntroductionDiamond is the cubic form of crystalline carbon. Carbon is the sixth element of the periodictable, with a proton number of 6 and atomic number of 12. With atomic number density of1.75 x 10 23 atoms/em", which corresponds to a molar volume of 3.44 cm 3/mol., the carbonatoms in the diamond lattice structure has thus one of the highest atomic number densities.80

The mass density ofdiamond is 3.515 g/cnr'. With these physical properties, diamonds, with asize ofa few cubic millimeters can be used in radiation measurements. The small size allowsfor 'pin point' measurements to be made which has an advantage over large volume detectorswhere average values are obtained. The lattice structure of diamond consists of singleelements of carbon in Sp3 bonded form i.e. in the stable four-atom coordination state. Siliconand germanium are elements in the same group as carbon in the Periodic Table of theElements. Silicon and germanium have the diamond structure and are also used for theconstruction ofwide range ofelectronic devices ofvaried applications hence forming the basisfor the comparisons between them and diamond. Table 5.1 is presented comparison ofsome ofthe properties ofdiamond and silicon.Table 5.1: Comparison ofsome ofthe properties ofdiamond and silicon.Property Diamond SiliconBand gap 5.47 1.12Mass density [g1cm3] 3.515 2.33Dielectric constant 5.7 11.9Resistivity [Wcm] >1011 2.3 x 105Breakdown field [kV/cm] 103 -2)\ IQ4 300Electron mobility [cm2Ns] 1500 - 2400 1450Hole mobility [cm2Ns] 1000 - 2100 480Thermal conductivity [WI em K] 10-20 1.3Radiation length [em] 12 9.4Energy to create an electron-hole pair [eV] I3 3.6References: Pan et al., 199381

The major setback to the application of natural diamond as a radiation detector has been thelengthy selection process for a good stone and the fact that natural diamond crystals differ onefrom another. Lightowlers and Dean (1965) noted that no two natural diamond crystals can beguaranteed to produce similar responses to radiation without extensive initial testing of eachstone. The development of synthetic diamond has created enthusiasm to the possiblecommercial exploitation of the crystal in radiation detection instruments. With the presentsynthesizing techniques, it is believed that batches of diamond crystals containing controlledamounts of impurities and having good detection characteristics can be produced in areproducible manner.The application of synthetic diamond crystal as dose measuringinstrument has since been reported. Diamond has been used as radiosensitive resistor(Burgerneister, 1981; Keddy et al., 1987); as a thermoluminescence dosimeter (Nam et al.,1987; Nam 1989); as a near tissue-equivalent probe in electron radiation therapy (van derMerwe, 1994); as a sensor for measuring low dose-rates (Grobbelaar et al., 1991).Fallon et a!. (1990) has found that as a pulse-counting gamma-ray detector, synthetic diamondhas a linear response for over five orders ofmagnitude ofdose. Fallon et a!. (1990) also foundthat priming diamond with a large y-ray dose gave best response. For a chemical vapourdeposited (CVD) diamond however, Bruzzi et a!. (2000) has found linearity for a range of I­II Gy/min. CVD diamond dosimeters have been found to have the highest dose sensitivitycompared to natural diamond and silicon diode dosimeters. CVD diamond was found toexhibit priming behaviour in which the output increases slowly with the delivered dose(Whitehead et al. 2001).82

5.2. Selection ofexposnre geometry for the diamond sensorsAn investigation into the exposure geometry of the diamond sensor that will producemaximum absorption ofthe incident X-ray beam was conducted in this part ofthe study. Theabsorption levels of two CVD diamond samples ofequal surface area but different thicknesseswere calculated for both edge-on' and flat-on radiation exposures. Mass attenuation coefficientdata as well as the densities of diamond and those of the metals used to form ohmic contactson the diamond samples where applicable were taken into consideration. The study wascarried out using spectral data measurements obtained at 25 and 30 kVps. For the calculations,both the absorption depths and absorption areas of the exposure geometries were taken intoconsideration.The percent total absorption for the entire energy spectrum as well absorption percent atdifferent segments of the energy spectrum (arbitrary chosen) were calculated for both sets ofspectral data. The percent of total photon flux incident on the diamond samples based on theexposure geometry as well as the percentage incident photon flux absorbed by the diamondsample were determined. The percentage absorption from the two geometries considered ispresented in Tables 5.2 and 5.3 for the 25 and 30 kVps spectral data respectively.For the two selected kVps, the percentage of the total photon flux incident on the diamondsamples in the edge-on geometry was found to be much lower due to the relatively smallerarea of the diamond sample available to the X-ray beam than that of the flat-on geometry. Itcan be concluded from the data that at the lower energy segment of the spectrum, theI In the flat-on exposure geometry impingingradiation is to the largestsurfaceareaofthe specimen but hassmaller depth.whereasin edge-on exposure geometry the impingingradiation is on the smallest surfaceareaof..the same specimen buthaving greaterdepth. Cf. Figures5.1 and 5.2.83

Table 5.2: Comparison ofabsorption level ofdiamond sample positioned in an edge-on geometry with theabsorption level in a flat-on geometry for exposure measurements, for a 25 nominal kVp x-ray spectrum.Diamond Absorption Absorption % Incident % Incident % Incident % photon flux % photon flux % photon flux % photon fluxexposure depth area photon flux photon flux photon nux absorbed by absorbed by absorbed by absorbed byGeometry (em) (em') in energy in energy in energy diamond in diamond in diamond in energy diamond in theinterval interval interval energy interval energy interval intervall.5-8 kcV In &.5-15 keY In 15.5-25 kcV In L5-8 keY In 8.5-15 keY In 15.5-25 keY Entire spectrumEDGE-ON I 0,05 21.96 1339 64,65 100 98,2 85.0 94.4EDGE-ON I 0, I 21.96 13,39 64,65 100 98.2 85,0 94.4FLAT-ON 0,05 1 21.96 13.39 64.65 93.5 31,4 14.7 46.5FLAT-ON 0.1 I 21.96 13.39 64.65 97,8 47.4 22.6 55.9iiTable.5.3: Comparison of absorption level of diamond sample positioned in an edge on geometry with theabsorption levelin a flat ongeometry forexposure measurements! for a 30 nominal kvp x-ray spectrum,Diamond Absorption Absorption % Incident % Incident % Incident % photon nux %phuton nux % photon flux % pholon fluxexposure depth area photon flux photon flux photon flux absorbed by absorbed by absorbed by absorbed byGeometry (em) (em') in energy in energy in energy diamond in diamond in diamond in energy diamond intheinterval interval interval energy interval energy interval intervalIn 1.5-8 keY In 8.5-15kcV In 15,5-30 keY In 1.5-8 keY In 8.5-15 keV In 15.5-30 kcV Entire spectrumEDGE-ON 1 0,05 19.36 J1.62 69,02 100 98.180,4 92.9EDGE-ON I 0.1 19.36 11.62 69.02 100 98.1 80A 92,9fLAT-ON 0.05 I 19.36 11.62 69,02 92,9 31.5 12,8 45.7FLAT-ON 0.1 1 19.36 11.62 69,02 97,6 47,3 20.0 55.0I

absorption levels of both exposure geometries are similar. At intermediate and higher energysegments of the spectrum however, the percentage incident photon flux absorbed by thediamond samples was much higher in the edge-on geometry than the flat-on geometry eventhough the incident photon flux was comparable in both geometries.Figure 5.1: Edge-on ~eometryFigure 5.2: Flat-on geometrytttttttIncident X-ray beam directiont t t t t tIncident X-ray beam directionThe edge-on geometry was found to be about four times more absorbing than the flat-ongeometry at the intermediate and higher energy segments considered for the study. Theobservation is in agreement with the fact that as X-ray energy increases, mass attenuationcoefficients of materials decreases hence larger thicknesses are required for appreciableattenuation ofthe X-ray beam. Mali et al. (2004) found similar results in their investigations.The results ofthis study indicated that generally, the edge-on exposure geometry has higherabsorption levels than the flat-on geometry and ought to be the X-ray exposure geometry ofchoice in radiation dose measurements especially in the manunography X-ray energy range.Because both edge-on and flat-on geometry have comparable absorption levels at lower X-ray85

energies, flat-on exposure geometry will be appropriate for use in lower X-ray beam energies(less than 10 keV) as a higher response for the same crystal size can be expected.5.3. Construction ofTest Probe (Mark I)A sketch of the diamond probe used for selection of diamond specimens is shown in Figure5.3. To provide ohmic contacts, the diamond sample (5) surfaces were metallized withtitanium, platinum and gold. These were evaporated under vacuum onto both surfaces of thesample. Care was taken throughout the metallization process ofthe diamond surfaces to ensurethat the edges of the diamond were not coated with any of the metals. This was foundnecessary to prevent electrical shorting and spark over. The diamond probe housing consistedof a Teflon (polytetrafluoroethylene, PTFE) (4) core. Electrical connections to externalelectronics were carried out through the use of a standard triaxial connector (I). Three carbonfibre rods (3) were used to provide contacts to the diamond. The central carbon fibre electrodeand the metallic disc shaped electrode that is connected on both sides with two carbon fibreelectrodes were used to complete the circuit. To reduce spurious noise, the outer aluminium(2) casing was earthed. For measurements, the probe was connected to a Wellhofer Dosimetrie(model CD 500) commercial electrometer. In all the measurements, the incident radiationbeam was perpendicular to the length of the probe. The probe was calibrated against PTWDiados secondary standard dosimeter.5.4. Sample preparationTwo single crystal and four polycrystalline synthetic diamond samples of volume rangingfrom 2.7 mm" to 100mm 3 were used for the study. The samples were mechanically polishedwith diamond paste of grain size I J.lIIl placed on a cast iron scaife and rotating at about 280086

evolutions per minute. The polished samples were boiled in 1:3:4 mixtures of nitric,perchloric and sulphuric acids for 30 minutes. The samples were further cleaned in a 1:4solution of "Contrad" (a surface cleaning reagent) and deionised water in order to remove anymetallic impurities as well as non-diamond carbon atoms. This was followed by tripleultrasonic rinsing with deionised water at different time settings (Makau and Derry, 2003).No. Description Material No. Descrionon MaterialI Triaxial connector Steel 5 Activesensingmaterial Diamond2 Detector outerhousing Aluminium 6 Probecover Aluminium3 Electrodes Carbon fibre rods 7 Samoledisc oositioner Aluminium4 Insulator Teflon 8 Flexible wire Insulated copper wireFigure 5.3: A constructed Test Probe with diamond as sensing material (Mark I).5.5. Sample characterizationIt has been established that the response of a diamond sample is dependent on the types ofimpurities and their levels or concentrations. Nitrogen is the most important impurity indiamond. Nitrogen in diamond determines the optical properties and also influences thermal,87

electrical as well as mechanical properties of diamond (Field, 1992). Nitrogen is also theprincipal factor responsible for many performance characteristics of diamond as a radiationdetector (Nam et al., 1991). Diamonds with high nitrogen concentration tend to have muchlower shallow trap concentrations than those synthesized with lower nitrogen levels (Nam,1989). Diamond crystals with high concentration of nitrogen have been found to exhibit lowspecific conductance (Narn, 1989). Nam also observed that diamonds with singlesubstitutional nitrogen have high recombination efficiency to the detriment ofresponse due tothe presence ofa dipole. With diamond as a TLD (thermoluminescence dosimeter), it has beenreported that about 5 orders of magnitude reduction in the response could occur as theparamagnetic nitrogen content increases from 10 ppm to 150 ppm (Keddy et aI., 1987). It istherefore imperative to characterize the diamond samples to obtain information about the levelof impurities especially nitrogen levels and consequently establish the material quality. Thefollowing standard characterization techniques for crystals were carried out on the samples:Raman spectroscopy, electron spin resonance (ESR) measurements, current-voltage (I-V)characteristic measurements, polarizing voltage, response with and without UVR and responsewith time after exposure.5.5.1. Raman spectroscopyThe diamond samples were analyzed using Raman spectroscopy to determine the type ofcarbon atoms present in the diamond films. Raman spectroscopy was carried out before andafter the samples were polished and cleaned. Spectra were acquired using the micro-Ramanattachment of a Jobin-Yvon T64000 Raman spectrometer operated in single spectrographmode, with the 514.5 urn line ofan argon-ion laser as excitation source. The 20x objective ofthe Raman microscope was used for observation of typical beam diameter on the sample of88

just over 1 micron. In the measurement, light from the argon-ion laser beam was focused ontothe diamond sample and the backscattered light from the sample collected. Five different spotswere selected on the surface of each sample: one in the centre and 4 near the comers of therectangular samples. The sampling depth of the beam was approximately 5 microns. TheRaman spectral calibration was done using the green emission line of a mercury lamp atposition 1122.47 cm- I .The results of the Raman measurements were fitted with a Lorentzian function to determinethe Raman peak position and peak width at half height for each measured spot. The Ramanline of all the diamond samples (before and after they were polished and cleaned) occurred at1331.6 ± 0.5 cm- I . The peak width at half height for the diamond samples ranged from 2.25 ±0.01 to 2.56 ± 0.01 em", The Raman line peak intensities of the cleaned samples were higherthan those of the uncleaned samples. This was proposed to be due to the removal ofgraphiticand other non-diamond atoms and enhancement of the diamond atoms during the polishingand cleaning process. The 1 st order Raman scattering for diamond in an Sp3 bonded form isknown to be 1332 cm- I and has a half-width of about 3 cm- l (Field, 1992). The measuredRaman peak positions of the diamond films used in this study compare favourably with theestablished values indicating that the diamond samples used for the study were of goodquality. The Raman peak positions and the half-width values measured at different spots of asample were found to be similar for the same sample. These similarities show that there waslittle variation within the crystal structure of a sample. Raman spectrum ofdiamond sample Ais shown in Figure 5.4.89

10OOO,-~--_--- ~ -.80006000~1]u 400020001000 !l00 1200 1300 1400Wavenumber (em"]1500 1600 1700Figure 5.4: Raman spectrum of diamond sample A.5.5.2. ESR measurementsNitrogen in synthetic diamond occurs as isolated atoms in substitutional lattice sites atconcentrations up to 500 ppm (Field, 1992). In the substitutional position, four of the fivevalence electrons of the nitrogen atom form tetrahedral bonds with four surrounding carbonatoms. The remaining lone electron forms a C-N antibonding orbital that is evenly distributedover the four surrounding bonds resulting in the paramagnetic nature ofthe diamond (Fallon,1989). Electron spin resonance measurements (ESR) were carried out on the diamond samplesto determine the concentration ofnitrogen. ESR techniques depend on the interaction betweenthe spin angular momentum of an unpaired electron and an external magnetic field. In thepresence of a magnetic field, the degeneracy of the spin states of the unpaired electron90

characterized by the quantum number m, = ±Yz is lifted and transitions between the spin levelsare induced by radiation ofthe appropriate frequency.The ESR measurements were carried out with a Bruker ESP300E ESR spectrometer, operatedat a frequency of 9 GHz. The diamond sample was placed in a cavity resonator and variablemagnetic field was applied perpendicularly to the microwave magnetic field. In the magneticfield, the electron spin levels are separated by an amount, which depends on the magneticfield. At equilibrium, there is a small difference in the populations ofthe two levels. When themagnetic field is adjusted so that the energy difference between the two spin levels equals theenergy of the microwave photons, transitions induced between the two levels occurs, givingrise to an ESR signal.The observation ofthe ESR signal relies on maintaining the population difference between thespin levels, which in turn depend on the relaxation times. For high nitrogen concentrationsthese relaxation times are short and ESR transitions are detected readily. For low nitrogenconcentrations, the relaxation times become long and the population difference between thespin levels become smaller and it becomes more difficult to observe the ESR transitions. Atechnique, known as the rapid scan technique, in which the magnetic field is scannedrepetitively through resonance at a rate comparable to the relaxation time, can be used todetect the resonance. For both the conventional and the rapid scan technique, the concentrationin any sample is determined by comparing it with a signal from a calibrated reference sample.A typical ESR spectrum ofthe high-pressure high-nitrogen (HPHT) diamond sample (SampleF) with high concentration of nitrogen is shown in Figure 5.5. The clearly visible and equally91

spaced triplets in the figure indicate that the applied magnetic field was parallel to the [100]direction ofthe sample. The triplets occur at 3409.73g, 3443.45g and 3478.25g and had a linewidth of~I.46g.--.f-f-I I I I, ,3400 3420 3440 3460 3480 3500[6]Figure 5.5: Typical ESR spectrum ofa high nitrogen diamond sample.The electron spin resonance measurements of the polycrystalline CVD diamond, Samples Aand B had nitrogen concentration of20 and 10 atomic ppb respectively. The two single crystalHPHT diamond samples, Suite 2 No.2 and Sample F, had nitrogen concentrations of 16.9 and130 atomic ppm respectively.5.6. Metallization ofsampleThe polished and cleaned diamond samples were metallized following the method establishedby Hayes and da Costa, 2003. Metal contacts were applied by thermally evaporating onto thediamond surface using an electron beam system, successive metals: titanium (200 A)/platinum92

(200 A)/gold (2000 A), in an ultra high vacuum chamber at a pressure of about 10- 7 torrs. Themetal in contact with the diamond generally determines the electrical properties ofthe contact(Pan et al., 1993). Titanium and other transition metals that form carbides produce ohmiccontact to diamond. The metallized diamond was annealed (Tachibana et al., 1992). Platinumwas deposited next followed by gold at the surface. The chemically inert nature of goldprevents oxidation from occurring and thus protects the contact from any chemical reactionwith oxygen. Post-deposition annealing ofthe samples was carried out in a ceramic furnace ina high vacuum environment ofabout 10-6 torrs. The annealing temperature was 500°C.5.7. Sample groupingOperationally for ease of use and wide usage, an ideal detector should have both highsensitivity to the incident radiation, an extended range of linearity of response, and relativelyfast response. This study has shown that both response and sensitivity are susceptible to thepresence of defects within the crystal. In addition some diamonds due to the presence oftrapping centers are light sensitive. Two forms oftrapping centres, shallow or deep traps, havebeen identified. The shallow traps are believed to occur at energy level of less than 1.55 eVand are believed to be responsible for the slight initial increase in count rate after the diamondhas been pre-irradiated with a high dose of y-radiation. The pre-irradiation has been found topopulate these traps and depopulation occurs during pulse counting by exposure to lightenergy ofat least 1.55 eV, or by heat even at room temperature (Fallon, 1990).The deep traps located at energy levels greater than 2.2 ± 0.1 eV are believed to be responsiblefor the large decrease in response of the diamond detector after exposure to light. Populatingof the deep traps has also been found to be responsible for the increase in response when93

diamonds, with low concentrations of single substitutional nitrogen, have been pre-irradiated(Fallon, 1992). The depopulation of the trapping levels however, occurs when the diamond isexposed to light resulting in a lightly populated state, which acts as a carrier trap withconsequent reduction ofpulse-counting response (Fallon, 1992).To reduce the effect of traps and increase the performance ofthe CVD diamond sensor, it hasbeen reported that diamond samples with low nitrogen concentrations should be irradiated orprimed by exposure to y-, X-rays or ultraviolet radiation and should not be exposed to anyother light afterwards (Burgemeister, 1982; Keddy et al., 1987; Fallon et al., 1992; Vaitkus etal., 1993). Beta radioactive sources such as strontium-90 have also been used to prime CVDdiamond samples (Borchi et al., 1998; Marinelli et a!., 2001). Irradiation of diamond withultraviolet radiation has been found to have the same effect as pre-irradiation with y-radiationsources (Fallon, 1992, Vaitkus et al., 1993).Each diamond sample metallized on opposite faces for electrical contacts was carefully testedin order to identify working stones amongst them for use as the radiation sensor inmammography X-ray beam dose measurements. The tests consisted of: response toprogressive X-ray irradiation or dose; response with applied voltage; linearity with X-ray tubeloading and linearity with X-ray tube energy or tube voltage. From the test results, 'good'stones (Sample A, F and Suite 2 no. 2) were identified and 'not so good' ones (Sample B)were irradiated with ultra violet radiation and also with strontium-90 radioactive source withthe view ofimproving their performance.94

5.7.1. Selection of polarizing voltageThe applied potential for optimal performance and to obviate the dependence ofthe proberesponse on variations in applied potential of the diamond probe was established. The probewas connected to a Wellhofer Dosimetrie (model CD 500) commercial electrometer. Theresponse of the probe at varying applied potentials was recorded for the different diamondsamples. Saturation in response occurred at polarizing potential range of +280 V to +480 V,for both the HPHT single crystal diamond (Suite 2 No.2) and the CVD polycrystalline ones(Samples A and B). The probe was operated at +340 V (6.8 kV/cm) for all subsequentmeasurements. Figure 5.6 shows a response versus applied electric field curve for diamondsample A.5.7.2. Detector response to tube loading and tube voltage variationsThe response ofthe diamond samples used for the study was each measured. The sampleperformance testing was to gather baseline information about the samples i.e. to assess theirbehaviour with dose or with progressive X-ray irradiation. In order to establish the optimumtime needed for the probe to collect all the charges produced, the response of the probe at aspecific X-ray tube setting was measured with different electrometer time settings. Theresponse of the diamond samples with varying nominal X-ray tube voltage at a constant tubecurrent time product (tube loading, mAs) was carried out to determine the linearity of thediamond probe with energy. The linearity ofthe probe with tube loading was also measured ata constant tube voltage. Graphs ofthe linearity measurements are presented in Figures 5.7 and5.8. The linearity of the diamond probe was also checked against the response of ICIOionization chamber (Figure 5.9).95

0.6,;~d~0.3u~c;0.08-~u0< -0.3-0.6-0.9-1.23~/456Applied electric field (kV/cm)Figure 5.6: Response of diamond sample A with applied electricfield at an X-ray tube selling of 30 kVp and 200 mAs ,7870T60~:i~ 50u~c0 c.~ 40~T30T2~4 25 2627 28 29 30 31 32 33X-ray tube voltage (kVp, nominal values)Figure 5.7: Response ofdiamond sample A with varying X-ray tube voltage.Tube loading setting was 200 mAs and applied electric field was 6.8 kv/cm.96

~:j12010080/~"~s 60"-0~: "402040 80 120 160 200 240 280 320 360 400mAs (nominal values)Figure 5.8: Response of diamond sample A with varying mAs.X-ray tube voltage used was 30 kVp. The electric field appliedto the diamond sample was 6.8 kV/cm.120100I801~:j~ 60T"~"0"-~~ 4020 /-0.5 1.0 1.5 2.0 2.5Response ofIelD ionizationchamber(cGy)3.0Figure 5.9: Correlation ofthe response of diamond sample A withthe response aflelD ionization chambersetat sametube settings.97

The response of the HPHT single crystal diamond (Sample F) and the CVD polycrystallinediamond (Sample A) with varying applied voltage, nominal X-ray tube voltage and tubeloading have been compared. The response with X-ray tube voltage was carried out at aconstant tube loading of 200 mAs whiles that with tube loading was done at a constant tubevoltage of 30 kVp. Graphs showing the performance comparisons are presented in Figures5.10, 5.11 and 5.12. The measurements were carried out using PTW-Unidos E electrometersystem. The samples showed similar response patterns with varying applied voltage, X-raytube voltage and tube loading (Figures 5.10, 5.11 and 5.12 respectively). The response ofSample A was however, found to be higher than that of Sample F by a factor of 1.3 to 1.5 withthe higher factor occurring at higher applied voltage.50 •40 •Sample~~" 30 • •~"~• •eSampleF8- •~20"••"'"10 ••••••0 0 50 100 150 200 250 300 350 400 450Applied voltage(V)Figure 5.10: Response of diamond sample A and F with varying appliedpotential usingPTW-Unidos Eelectrometer system. Xeray tube settingusedformeasurement \V35 30 kVpand 200 mAs.98

6055•"~-e,5045&40~0iFc 35'"3025•••••••SampleA•Sample F•••20 24 26 28 30 32 34kVp(nominal values)Figure 5.11: Response of diamond sample A and F with varying X-ray tube voltage. X-raytube loading usedformeasurement was200 mAsandapplied voltage tothe diamond sampleswas400V. Measurements were carried outwith PTW-UnidosEelectrometer system50-e, " 40&§iF•Q30'" • •20 •·Sample A• SampleF••••mAs (nominal values)Figure 5.12:Response of diamond sampleA andF with varying X-ray tube loading. X-raytube voltage used formeasurement was 30 kVpandapplied voltageto thediamond sampleswas 400 V. Measurements werecarried outwith Pf'W-Unidos E electrometer.99

5.8. Priming ofdiamond samplesThe sensitivity of one of the diamond samples (Sample B) was found to increase withprogressive X-ray irradiation (Figure 5.13). This observation was assumed to be due to thefilling of trapping centres within the crystal by charge carriers. To test the 'trapping fillinggives rise to higher sensitivity to radiation exposure' hypothesis, the diamond sample wasirradiated with ultra violet radiation (UVR) for two hours. After the irradiation, the samplewas shielded from direct exposure to light. An increase in the response values of the sampleafter the UVR irradiation (Figures 5.14) with much lower applied voltage needed formeasurements was observed. The response of such sample was monitored for some time inorder to assess the temporal efficacy ofthe priming. The response was stable up to a point200 De180160""140C"~- c.=i 120~ ~.S0.>~ 100.:- B o 0 (,0" c, .: (J o o~80 c. c0 o oo.>c: oCG 0c:604020i>.~"'$'A0 000100 120 140 160 180Appliedvoltage (V)Figure5.13:Sensitivity of diamond sample B withprogressive X-rayirradiation.A, B, C andD represent the first,second, thirdand fourth measurements respectively.100

180160140120F:j 100F40200 0 30~00 "c0to.8000c< "60rAfterUVRI BeforeUVR60 90120 150Applied voltage (V)Figure5.14: Comparison ofthe response of diamond sampleB beforeand after UVRirradiationwith applied voltage.The X-raytube settingsusedwere30 kY, 200 mAs.(Figure 5.15) and thereafter showed an increase in response with dose indicating that not allthe trapping centres in the sample were filled, suggesting therefore that the 2-hour UVR wasinsufficient and that a longer irradiation time was needed.To further test the hypothesis that the diamond sample contains trapping centres the presenceof which gave rise to reduced response, the sample was annealed at 450°C for 30 minutes toempty as many of the shallow to medium level traps as possible. The diamond samplesshowed lower response values (Figure 5.16) compared to the response values before annealing(Figure 5.16). Much higher voltages could be applied to the diamond samples for a fixedexposure. The observation supports the results made after UVR irradiation. The response ofthe sample increased with accumulated dose (Figure 5.17) needing consequently a lower101

voltage for the same response to exposure. The response ofthe diamond sample did not showinitial drop, contrary to the response values before the UVR irradiation. This observationsuggests that the annealing process adopted did not empty all the filled traps during theheating cycle implying that only partial annealing ofthe sample was achieved.Furthermore where an annealed diamond sample was irradiated with UVR for 5 hours, acomparison of the response of the diamond samples from the two UVR irradiation times(Figure 5.18) with a fixed applied voltage showed that the 2-hour UVR irradiation gave higherresponse values than the 5-hour UVR irradiation. The observation suggests that the UVR hasnot been successful in improving the performance of the diamond sample. The decay times(Figure 5.19) ofthe sample after irradiation also confirm the presence oftraps.46442: 42~cc~40~] 38a.0-< 363432Accumulated dose (cGy)Figure 5.15:Variation of appliedvoltagerequired for fixed response with accumulated dose values.Sarrple has been UVRirradiated for2 hours. TheX-ray tubesettings used"ere 30kv. 200mxs.102

180160140120~~100~~~c 80a 0.BeforeannealingAfterannealing"~~60c40200 24 25 26 27 28 29 30 31 32X-raytube voltage (nominal values)Figure5.16:Comparison of the response of diamond sampleB beforeand afterannealing at 450 C. Beforeannealing measurements werecarried out at 32 Vwhilethe afterannealing measurements weredone at 300 V.300 .c 2.~ 250~~~ 200§:

The presence of traps in the diamond detector has been found to give rise to slow decay timeafter a pulsed excitation with a consequent rise in baseline level (Bergonzo et al., 2003). Thediamond sample was further irradiated with strontium-90 source for 60 hours (Marinelli et aI.,2001). It was expected that because of its higher intensities and energies the strontium-90source would be able to penetrate the entire diamond thickness with increased probability oftrap filling in a shorter time as compared to UVR. A fully primed sample was expected tofunction more efficiently than previously observed. The diamond sample response after thestrontium-90 irradiation was higher than before the strontium-90 irradiation. However, likeearlier observations made in previous measurements, the response of the sample increasedwith accumulated dose.180160140120~=! 100$"~c: 80 00-~"'" 604020o 0oo 2hoors UVRc oo0 05 hours UVRo 0oo00 00020406080100120140X-ray tube voltage (nominal values)5.18:~ ofthe response ofdiarrondsample B irradiated at differenttimes with applied voltage, The X-raytubesettings used "ere30 kV, 200mAs.104

22020000180 0 00160oi~ 140"~~ 120~"'" 10008060-1 0 2 3 4 5 6Time (min.) afterX-ray irradiationFigure5.19: Decaytime of diamondsampleB afterexposure.Response at zero represents beforeirradiation level.It is apparent from the results above that UVR and strontium-90 irradiations which are usuallyused to prime low nitrogen CVD diamonds have not been successful in improving theperformance of this particular diamond stone. Other forms of sample passivation techniqueshave to be considered and employed. This however, is beyond the scope ofthe present study.5.9. Construction of diamond probeFollowing the performance of the constructed Test Probe (Mark I), two more diamond probeshave been constructed with similar electrical connections as in Mark I. The two probes namedas Mark IIA and lIB (Figures 5.20 and 5.21 respectively) were constructed using entirelytissue equivalent materials with carbon fibre providing both electrical contacts and shielding.Electrical insulation was provided by Teflon tubes. The outer casing ofthe probes was105

No, Descnmion Materia! No Descrintion Material1 Triaxial connector Steel 7 Insulated holders Rubber2 Probe Persnex 8 Detector holder PCn'>I)CX3 Detectorouter housing Perspex 9 Detector cover Persoex4 Insulated wires Copper to Activesensina materia! CVD Diamond5 Rodconductors Conner 11 Split halvesholding rin. PcrsnexIi Electrodes Carbon fibre rods I~_M1.5J,.~1'.1~1"ttlma..fTlIDI,.~la.T-HlS seCTION WITH'rOP 8f't.ITtW.Jl Rr;MOV£Ilill II41 81 !~ICI~I ~ill il•13 __ ,_ 12,5:lOA ...2

No. Descrintion Material No. Descrintion Material1 Triaxial connector Steel 7 Insulated holders Rubber2 Probe Persnex 8 Detector holder PcrsnexJ Detector outer housina Persncx 9 Detector COver Persncx4 Insulated wires Conner 10 Active scnsin;;-material CVDDiamond5 Rod conductors Conner J I Split halves holdina rina Persnex6 Electrodes Carbon fibrerods~."'1.t5,~12, nwa.__ 15,loc:.a.d 7 toma..THIS8EC11ON WITHTOP SPUT HAU' REMO'IeD~11~ il;[~~1;f/>6/t/Bz6z®z6"213 I 12.120,0: 21". •1.5 13~ I'..222All dimensions InmmFigure 5.21: CVD diamond probe with Perspex casing, Mark JIO.o......

Perspex and it has removable split Perspex cap or window, which allows for the directmeasurements ofincident photons. The outer casing ofthe probes was earthed with 3 micronsthickness of aluminium in order to reduce electronic noise. Mark IIA (Figure 5.20) aredesigned for diamond stone of volume 52 mrrr' (10 x 10 x 0.52 rnrrr') (Sample A) and isintended for total dose measurements while Mark lIB (Figure 5.21) with a crystal of volume2.7 mnr' (Suite 2 no. 2) is targeted for routine dose monitoring. Mark IIA has been constructedto allow for both edge-on and flat-on exposure measurements. Photographs of Mark IIA areshown in Figures 5.22 and 5.23.Some salient features of and anticipated advantages that could artse from the speciallydesigned and constructed diamond probes and with preselected crystals are:(1) Dose response values are not susceptible to changes in ambient temperature, pressure andhumidity unlike air ionization chambers. (2) Diamond is near tissue-equivalent thus it hassimilar attenuation properties as tissue therefore its response in dose measurements would notrequire some of the conversion factors needed to convert air kerma measurements to meanglandular dose values. The number of variables used in dose evaluation quantities is reducedhence reduction of possible sources of error or uncertainty in dose measurements. (3) Highsensitivity of diamond implies that thinner diamond samples could be used and hencereduction in total cost ofconstruction. (4) With its customized Perspex phantom, tissue depthdose can easily be measured. (5) The constructed probes are easily adaptable to existingcommercial electrometers and thus do not need special electrometer systems for operation. (6)The construction design of the probes allow for radiation detection in both "edge-on" and~~flat-on" sensor geometry profiles without firstly unseating the diamond sensor element fromits original position within the probe housing before taking measurements. This feature makes108

Figure 5.21 : Photograph of Mark UA.Fizure 5.23: Photograph of Mark UA in its customized Perspex phanrom.- -109

the probe physically more robust than the currently available ionization chambers. (7) Theconstructed probe is expected to find use for charged particles detection in the flat-ongeometry and therefore may find application in electron and proton therapy as well. The probewith diamond as a sensing material thus has a further possibility of being able to replace thedifferent ionization probes, currently needed to provide dose evaluation information.5.10. Dose measurementsTo determine radiation dose in Perspex material and hence the dose to glandular tissue, theconstructed probe was inserted into the custom-made Perspex phantom. Varying thicknessesofPerspex were placed on top ofthe Perspex phantom and each time X-ray tube exposure wastaken and the diamond probe response recorded. Exposure of the diamond probe with zeroPerspex was also taken. The difference between the response with and without Perspex wascalculated as the radiation dose to the Perspex material. In this measurement the diamondsensor was in "edge-on" geometry configuration. The calculated radiation dose for varyingthicknesses of Perspex from the measurement has been plotted and compared with calculatedradiation dose obtained from simulated studies. The results are presented in Chapter 6.5.11. Effect of a breast compression plate on patient doseThe extent to which breast compression plate influences the radiation dose to patient has beenassessed in this part ofthe study. Response ofthe diamond sensor to direct incident ofphotonswas carried out with the Perspex cap or window ofthe probe removed. Varying thicknesses ofPerspex were placed on top ofthe diamond; exposures were taken and the responses recorded.The assessment was conducted for different X-ray tube settings. It was observed that the 3 mmPerspex thickness commonly used as breast compression plate in mammography X-ray110

examination reduces radiation dose to the patient significantly. For a 25 kVp examination, thereduction in radiation dose was found to be 23%. The radiation dose reduction for 26, 28 and30 kVp examinations were found to be 21%, 16% and 14% respectively. The result from theevaluation is presented in Figure 5.24. The results obtained highlights the often forgotten rolethat a breast compression plate has in reducing the dose to patients and the degree to which itis reduced as a function ofthe tube voltage setting.504530kVp•403528kVp~•••"~30~§~~ 25""2015• 26kVp•• ••25 kVp•••••10o 23Perspex thickness (mm)5Figure 5.24: Response with small increment ofPerspex thickness for differenttube voltage settings. A tube loading of 200 mAs was used in all the measurements.Graph demonstrates the importance of breast compression plate in dose reduction.ReferencesBergonzo P., Tromson D., and Mer C. (2003). Radiation detection devices made from CVDdiamond. Semicond. Sci.Techno!. 18 (3), SI05-S112.Borchi E., Bruzzi M., Leroy C., and Sciortino S. (1998). Thermoluminescence analysis of 13-and y-irradiated chemical vapour deposited diamond films. J. Phys. D.: Appl. Phys. 31,609-616.III

Bruzzi M., Bucciolini M., Cirrone G.A.P., Cuttone G., Mazzocchi S., Pirollo S., and SciortinoS. (2000). Characterization of CVD diamond dosimeters in on-line configuration.Nuc!. Instrum. Methods Phys. Res. A 454,142-146.Burgemeister E.A. (1981). Dosimetry with a diamond operating as a resistor. Phys. Med. Bio!.26 (2),269-275.Burgemeister E.A. (1982). Settling times of radiation detectors made from diamonds.Proc. 8 th Symp. On Microdosimetry, Kemforschungsanlage Julich GmbH Julich, WestGermany, 27 September to 1 October 1982 (Brussels: Commission of the EuropeanCommunities), 993-1001.Fallon P.J., Nam T.L., Keddy RJ., Bums R.C. and Grobbelaar lH. (1990). Synthetic diamondused as pulse-counting y-ray detectors. App!. Radiat Isot. 41, 35.Fallon P.J., Nam T.L., and Keddy R.J. (1992). Trapping levels in pulse-counting syntheticdiamond detectors. Diamond and Related Materials 1, 1185-1189.Fallon P.J. (1989). Synthetic diamonds as pulse counting radiation detectors. MSc thesis,Faculty ofScience, University ofthe Witwatersrand, Johannesburg, South Africa. 62.Field J.E. (editor) (1992). The properties of natural and synthetic diamond, Academic PressLimited, 67-71.Grobbelaar J.H., Burns R.C., Nam T.!. and Keddy R.J. (1991). Miniaturized radiationdetector with custom synthesized diamond crystal as sensor. Nuc!. Instrum. MethodsPhys. Res. B61, 553-559.Hayes M. and da Costa A.M.O.D. (2003). Private communication.Keddy RJ., Nam T.L. and Burns R.e. (1987). Synthetic diamonds as ionization chamberradiation detectors in biological environments. Phys. Med. Bio!. 32 (6), 751-759.112

Lightowlers E.C. and Dean P.J. (1965). An efficient method for selecting type 11andIntermediate-type diamonds. Ind. Diamond Rev. 25,143-146.Makau N.W. and Derry T.E. (2003). Study ofoxygen on the three low index diamond surfacesby XPS. Surface Review and Letters 10 (2 & 3), 295-301.Mali T, Cindro V., Mikui M., Zdesar U., and Jancar B. (2004). Evaluation of siliconmicrostrip detectors as x-ray sensors in digital mammography. Article from theInternet. Jozef Stefan Institute, Jamova 39 SI-IOOO Ljubljana, Slovenia(corresponding author's address).Marinelli M., Milani E., Paoletti A., Tucciarone A., Verona Rinati G., Angelone M., andPillon M. (200 I). Systematic study ofthe normal and pumped state of high efficiencydiamond particle detectors grown by chemical vapour deposition. J. App!. Phys. 89 (2),1430-1435.Nam T.L., Keddy RJ. and Bums R.C. (1987). Synthetic diamonds as III VIVO radiationdetectors. Med. Phys. 14 (4), 596.Nam TL. (1989). Nuclear radiation detection properties of diamond. PhD thesis, Faculty ofScience, University ofthe Witwatersrand, Johannesburg, South Africa, 72-77.Nam TL., Karfunkel U., Keddy RJ., and Every A.G. (1991). The effect of nitrogen impurityon the radiation detection properties of synthetic diamond. Radiation effect andDefects in Solids 116,233-252.Pan L.S., Han S., Kania D.R., Zhao S., Gan K.K., Kagan H., Kass R., Malchow R., MorrowF., Palmer W.F., White C., Kim S.K., Sannes F., Schnetzer S., Stone R., ThomsonG.B., Sugimoto Y., Fry A., Kanda S., Olsen S., Franklin M., Ager 1. W., and PianettaP. (1993). Particle- and photoinduced conductivity in type-lla diamonds. 1. App!.Phys. 74 (2),1086-1095.113

Tachibana 1., Williams RE., and Glass J.T. (1992). Correlation of the electrical propertiesof metal contacts on diamond films with the chemical nature of the metal-diamondinterfaces. II. Titanium contacts: a carbide-forming metal. Phys. Rev. B 45 (20),11975-11981.Vaitkus R., Inushirna T., and Yamazaki S. (1993). Enhancement of photosensitivity byultraviolet irradiation and photoconductivity spectra of diamond thin films. Appl.Phys. Lett. 62 (19), 2384-2386.van der Merwe D.G. (1994). The effect of tissue inhomogeneities on the energy spectrumand dosimetry in electron radiation therapy. PhD thesis, Faculty of Science,University ofthe Witwatersrand, Johannesburg, South Africa.Whitehead A.J., Airey R., Buttar C.M., Conway J., Hill G., Rarnkumar S., Scarsbrook G.A.,Sussmann R.S., and Walker S. (200!). CVD diamond for medical dosimetryapplications. Nuc!. Instrum. Methods Phys. Res. A 460, 20-26.114

Chapter 6COMPARATIVE STUDIES OF MEASUREDMAMMOGRAPHY X-RAY BEAM DOSES WITHTHEORETICAL CALCULATION FROMMONTE CARLO CODE SYSTEM PENELOPEUSING SYNTHETIC DIAMOND ASRADIATION SENSOR115

AbstractThe utilization of a probe with synthetic diamond as a sensing material to measure radiationdoses from a mammography X-ray beam was carried out. A computer code systemPENELOPE that simulates coupled electron-photon transport has been used to computeradiation doses from mammography X-ray beams to the diamond sensing material. Thesimulation algorithm is based on a scattering model that combines numerical databases withanalytical cross section models for different interaction mechanisms. Investigation into theexposure geometry of the diamond sensor that would provide maximum absorption of theincident X-ray beam has also been conducted using PENELOPE code. A model wasdeveloped that simulates the experimental setup and conditions for evaluating mammographyX-ray beam radiation doses using diamond as the conduction sensing element. The resultsfrom the theoretical model and experimental measurements are compared and presented intabular and graphical forms.6.1. IntroductionMonte Carlo (MC) methods are known to be exact and accurate to the degree to whichphysical parameters are known and with sufficient statistics have superior accuracy overdeterministic and analytical methods. Some of the powerful and well-established MC codesare: EGS4 - Electron Gamma Shower version 4 (Nelson et al, 1985; Bielajew et al., 1994);EGSnrc extended and improved version ofEGS4 for calculating the response of ion chambersused in medical physics; ETRAN - Electron transport code and ITS - Integrated Tiger Series(Halbleib, 1989; Halbleib et al., 1992); GEAT MC simulation tool originally developed atCERN for high energy physics experiments; MCNP - Monte Carlo N-particie transport code116

(Briesmeister, 1993) and PENELOPE (Baro et al., 1995; Salvat et al. 1996; Sempau et al.1997). These MC codes have been benchmarked against experimental data and found to be ingood agreement for a wide range of energies and materials. MC simulation of radiationtransport is generally accepted to be the most accurate means of predicting dose distributionsespecially in the presence of sharp heterogeneities such as soft tissue, bone, air cavities andlung. MC codes have been extensively used in radiation treatment planning of patients(Walling, 2000; Li et aI., 2000; Lalic et al., 2001; Sempau et al., 2001; Rodriguez et aI., 2002;Cherty et aI., 2003; Siegbahn et al. 2003).Mammography X-ray beam absorbed dose has also been assessed with Monte Carlosimulations. Doi and Heang-ping in 1980 evaluated mammography absorbed dose using MCsimulation. Dance (1980) also calculated integral radiation dose in xeromammography withMC code. Some investigations on mammography techniques using MC code include: MCcalculation of conversion factors for the estimation of mean glandular breast dose by Dance(1990); backscatter factors for mammography calculated with MC methods by Kramer et aI.,(2001); MC simulation ofX-ray spectra generated by kilo-electron-volt electrons by Llovet etal. (2003).In the present work, PENELOPE (PENetration and Energy Loss of Positrons and Electrons), aMC algorithm and computer code designed specifically for the simulation ofcoupled electronphotontransport in arbitrary materials is used to assess absorbed dose from mammography X-.ray beam. PENELOPE was developed by Salvat et al. (2001) with the aim of providing ageneral-purpose MC code with enhanced transport algorithms. The first package of the code117

was released in 1996. The simulation algorithm ofPENELOPE is based on a scattering modelthat combines numerical databases with analytical cross section models for differentinteraction mechanisms and is applicable to energies from a few hundred eV to - 1 GeV. Thecode performs "analogue" simulations i.e, simulated showers are intended to be replicas ofactual showers of electron-photon simulations in infinite media of various compositions. Thereliability of PENELOPE in low-energy photon dose calculations has been demonstrated bycomparison studies with other MC codes and also with experimental data and showedexcellent agreements (Bare et al., 1994; Baro et a!., 1995; Sempau et a!., 1997; Ye et a!.,2004). The detailed characteristics ofPENELOPE code is reported by Baro et a!. (1995).6.2. Simulation of photon transportPhoton interaction mechanisms assumed III the code are those of coherent or Rayleighscattering, incoherent or Compton scattering, photoelectric absorption and positron-electronpair production. The orbital binding effects and Doppler broadening are taken into account inthe Compton scattering cross sections. In simulating photoelectric absorption, thecharacteristic K-shell X-rays and Auger electron emissions are accounted for. Photon transportis simulated by the conventional detailed method in which all interaction events experiencedby a particle is simulated in chronological succession. This is in view ofthe fact that photonsundergo a limited number of interactions before they are locally absorbed thus requiringrecomputation ofthe distance to collision at every media boundary.The accuracy ofMC simulations strongly depends on the accuracy in the probability118

density functions and thus on the cross section libraries used for photon-transport calculations.PENELOPE Me code uses photoelectric interaction cross sections from the most updatedEPDL97 library (Evaluated Photon Data Library) housed by LLNL (Lawrence LivermoreNational Laboratory). The data has been calculated to higher precision to allow for moreaccurate interpolation between tabulated data points. With the updated cross sections andenhanced transport algorithms, PENELOPE has been shown to have excellent agreement inlow-energy photon dose calculations (Ye et a1., 2004).6.3. Implementation ofPENELOPE codeThe code was installed and run on a 1600 MHz Pentium 4 Intel processor and 256 MB RAMpersonal computer. PENELOPE is implemented in a FORTRAN 77 computer code and runson any platform with a FORTRAN 77 or FORTRAN 90 compiler. The source files ofPENELOPE code system includes: the transport and physics routines (penelope.f); the quadricgeometry package (pengeom.f); variance-reduction routines (penvared.f); the main program tocreate cross-section data files (material.f); and a program which generates tables of materialinteraction properties (tables.f). There are other packages in PENELOPE such as material.exefor creating material cross section data; GNUPLOT for plotting simulated results; timer.f asubroutine used in timing the simulation process and GVIEW a software for geometryvisualization.The source file of the code PENELOPE.F consists of four mam blocks of subprograms,namely: the preparatory calculations and input and output (I/O) routines, interactionsimulation procedures, numerical routines and transport routines. The PENELOPE code119

simulates these interactions though analytical differential cross sections (DCSs) derived fromsimple physical models. To minimize loss of accuracy as a result of using analytical DCSsinstead of numerical data, physically plausible analytical forms are adopted in the code. Theinteraction results are also renormalized by the code to reproduce values ofpartial attenuationcoefficients that are read from the input material data file.The code is applicable to both single medium and multimedia environment such as mixturesand compounds. In order to simulate photon interactions with compounds and mixturesadditivity approximation methods are adopted to obtain the DCSs ofthe compound or mixtureby summing the corresponding DCSs ofall the atoms in the compound or mixture of interest.To use PENELOPE code the executables files needed to run the code have to be compiled andlinked as indicated by accompanying instructions (Salvat et aI., 200 I).6.4. Input data files needed to run PENELOPESimulation of arbitrary geometry and scoring using PENELOPE is possible without previousknowledge of the intricate theoretical aspects of scattering and transport theories. However,the user of the code is required to provide the steering programs, which controls the geometryand evolution of tracks, keeps score of the relevant quantities and performs the requiredaverages at the end ofthe simulation. The input data files that must be supplied in order to runthe code include: (I) geometry definition file; (2) material data file and (3) the input file.120

6.4.1 Geometry definition fileSimulations of radiation transport in material systems involve two different kinds ofoperations namely: physical and geometrical. The physical, determines the path length to thenext interaction and the random sampling ofthe different interactions. The geometry involvesspace displacements, interface crossings, identification of bodies and materials present atinterfaces. The geometry definition file is an input file which consists of a series of data sets(strictly formatted text lines), which defme the different elements surfaces', bodies 2andmodules. The geometry definition file is interpreted by the subroutine PENGEOM. Toevaluate mammography X-ray beam radiation doses using diamond as a sensing material, ageometry definition file for the constructed prototype diamond probe was written for thePENELOPE code to simulate the experimental setup and conditions for measurements. Thedimensions of the Perspex and diamond including those of the contacts materials (whereapplicable), were taken into consideration as well their physical orientations. A 3D display ofthe geometry files for the edge-on and flat-on geometries are shown in Figures 6.1 and 6.2.6.4.2. Material data fileMaterial data file is the section of PENELOPE code where information about each materialsuch as tables of physical properties, interaction cross sections, relaxation data, densities andother physical constants are stored and read. The auxiliary program MATERIAL was used tocreate the material data file. The material data file for each ofthe materials used in the1 Surfaces are assumed to be quadrics and include: planes, pairs ofplanes, spheres and cones. Others are:cylinders, ellipsoids, paraboloids and hyperboloids.2 A body is a space volume limited by quadric surfaces and filled with a homogeneous material while a module isa connected volume, which contains one or several bodies and is limited by quadric surfaces.Surfaces, bodies and modules can be rotated using Euler angles defined as in Edmonds (1960).121

Figure 6.1: The geometry diagram of the diamond probe in the edge-onexposure geometry used for the simulation studies with a sector excludedto show the geometry of the interior (diamond location).Figure 6.2: The geometry diagram of the diamond probe in the flat-onexposure geometry used for the simulation studies with a sectorexcluded to show the geometry of the interior (diamond location).122

construction of the probe was supplied in response to prompts from the program. This wasthen concatenated starting from the filename ofthe outer material ofthe probe to the inner oneto create a single material data file for the prototype diamond probe.6.4.2. The input fileThe distribution package of PENELOPE code includes various examples of MAIN programs.To operate any of the MAIN programs, an input data file has to be provided. The input filesare written according to specific order and format with an explicit 6-character lkeywordfollowed by numerical data or a character string. The 6-character keywords identify thesimulation parameters such as the kind of particle being simulated; the energy, and positionfrom target; and its angular aperture described by the polar and azimuthal angles with respectto the Z-axis, the default axis vector direction for any beam; the number of materials in thegeometry being simulated; the material and the geometry filenames. Other parameters alsoindicated in the input file include the cut-off absorption energies (for each material), thedesired number of simulated showers and the simulation time. The keywords and numericaldata supplied are verified by the program.In this work, the MAIN program PENDOSES was used to simulate the absorbed dose in theprototype diamond probe. It was also used for the exposure geometry of maximum absorptionevaluation exercise for the selection of the edge-on or flat-on configurations as exposuregeometry ofchoice. Whilst the program is distributed for monoenergetic particles the programI Ifthe keyword is less than 6 characters, blank space denoted by "-" is used. The numerical data starts at the gilicolumn.123

wasmodified to accept spectra energy distribution ill order to simulate experimentalconditions correctly.6.5. Selection of exposure geometry for the diamond sensors using PENELOPEAn investigation into the exposure geometry of the diamond sensor to give maximumabsorption ofthe incident X-ray beam was simulated using PENELOPE. The absorption levelsof two diamond samples of equal surface areas but different thicknesses were calculated forboth edge-on and flat-on radiation exposures. The geometry files representing the edge-on andflat-on geometries and which are required by the Code to describe the geometry of thesimulation were written (Figures 6.1 and 6.2) for both diamond sizes and used to simulateexposure geometries independently. Monoenergetic beams at 25 keY as well as a typical 25-kVp molybdenum energy spectrum were employed for the simulation. For the latter, 500 eVenergy bins covering the entire spectrum were used.6.5.1. Interpretation of simulated results for the geometry selection studyAt each energy, over I x 10 6 primary particles were simulated resulting in statistical errors ofless than I%. The results from the simulation were analyzed to obtain the dose deposited inthe diamond. To obtain a figure of merit for the results from the simulation using differentbeam apertures, the number of impinging particles N, onto the medium was first correctedusing the expression:N = Nabs xN f.C Na(6.1)124

The parameters Nabs, Na and N] are the number of absorbed primary particles by the medium,the number of impinging particles at the angle of interest and the number of impingingparticles at the full angle Isubtended by the beam onto the medium respectively, The totalenergy deposited E"d in the diamond is determined from the corrected number ofparticles,where Ed is the deposited energy per simulated primary particle. The following observationswere highlighted by the simulation study:(6.2)6.5.1.1. Edge-on geometryThe simulated number ofparticles was higher for an incident monoenergetic beam than for anincident energy spectrum (Table 6.1). This is due to the fact that impinging particles all havean energy of 25 keY in contrast to an energy spectrum where the particles have, not onlyranging energies (energy ranged from 2 keV to 25 keV), but also ranging probabilitydistribution functions all ofwhich have to be taken into account during the computation; manymore computations therefore per particle compared to the monoenergetic beam.Thefractional particle absorption is higher in the case of the energy spectrum as compared to themonoenergetic beam; the energy spectrum contains particles with lower energies havinghigher absorption cross sections whereas the monoenergetic beam has no lower energycomponents.The fractional particle absorption of the energy spectrum is calculated to beabout 3x higher than for the monoenergetic beam resulting in the total energy deposited beingequally, three times greater. However, the energy deposited per particle in the case of theI The full angle e subtended by the beam is calculated from the length of the beam-facing medium I andimpinging particle to medium distance f, as: e ~2Ian-J(tl2}). The area in em' subtended by the beam onto themedium would be: A, ~ " ifx Ian (12)'.125

energy spectrum is only about I.2x greater than the monoenergetic beam. This, as expected,indicates that a significant percentage of the fractional particle absorption is due to the lowerenergy component of the spectrum but that the contribution to the overall energy deposited issmall. In both the monoenergetic and energy spectrum results, there is a decrease in depositedenergy per emitted particle and total deposited energy with increasing beam aperture. This isdue to the fact that the area projected by the selected beam apertures, except for cone angles0.01175° and 0.175°, at the diamond extended beyond the physical diamond crystal areairradiated.6.5.1.2. Flat-on geometryLike the edge-on geometry, the simulated number ofparticles was higher in the monoenergeticbeam (Table 6.2) than the energy spectrum but, again, the fractional particle absorption washigher in the energy spectrum than the monoenergetic beam. The fractional particle absorptionand the total energy deposited by the energy spectrum simulation were found to be about IIxand 8x higher respectively compared to the monoenergetic beam. The deposited energy perparticle was however, only 3x higher for the energy spectrum compared to the monoenergeticbeam. A similar observation was made in the edge-on geometry. While the fractional particleabsorption and deposited energies decreased inversely with beam aperture selected in theedge-on geometry, they were unchanged in the case of the flat-on geometry for reasonsalready described in the edge-on situation.126

Table 6.1: Comparison oftotal deposited energies in diamond using monoenergetic beam and energy spectrum inthe edge-on geometry as a function ofbeam aperture for a diamond with an exposure area of0.1 crrr' and athickness of 1 mm. Impinging monoenergetic photon employed was 25 keY.Monoenergetic beamBeam aperture Simulated Absorbedprimary Fractional particle Deposited energy Total depositedin degrees primary Particles absorption (keV) per primary energy (keV)Particlesparticle0.01175 2.326E+06 5.923E+05 0.255 6.802 4.204E+060.175 2.427E+06 6.175E+05 0.254 6.797 4.197E+060.5 3.524E+06 6.169E+05 0.175 4.675 1.986E+061.43 9.384£+06 6.061£+05 0.065 1.724 2.702E+052.86 1.835E+07 5.879£+05 0.032 0.856 6.656E+04Energy spectrumBeam aperture Simulated Absorbedprimary Fractional particle Deposited energy Total depositedin degrees primary Particles absorption (keV) per primary energy (keV)Particlesparticle0.01l75 2.082E+06 1.529E+06 0.735 8.312 1.330E+070.175 2.178E+06 1.600E+06 0.735 8.307 1.329E+070.5 3.098E+06 1.602E+06 0.517 5.795 6.627E+061.43 8.242E+06 1.576E+06 0.191 2.141 8.917E+052.86 1.554E+07 1.487E+06 0.096 1.068 2.226E+05127

Table 6.2: Comparison oftotal deposited energies in diamond usingmonoenergetic beam and energy spectrum inthe flat-on geometry as a function of beam aperture for a diamond with an exposure area of 1 crrr' anda thicknessof Imm. Impinging monoenergetic photon employed was 25 keY.Monoenergetic beamBeam aperture Simulated Absorbedprimary Fractional particle Deposited energy Total depositedin degrees primary Particles absorption (keV) per primary energy (keV)Particlesparticle0.01175 1.424E+07 5.527E+05 0.039 1.036 5.847E+050.175 1.454E+07 5.640E+05 0.039 1.036 5.843H050.5 1.454E+07 5.641H05 0.039 1.036 5.844E+051.43 1.455E+07 5.642H05 0.039 1.036 5.845E+052.86 1.463H07 5.631H05 0.038 1.028 5.753E+05Energy spectromBeam aperture Simulated Absorbedprimary Fractional particle Deposited energy Total depositedin degrees primary Particles absorption (keV) per primary energy (keV)Particlesparticle0.01175 3.43IE+06 1.406E+06 0.410 3.209 4.513E+060.175 3.432E+06 1.406E+06 0.410 3.209 4.512E+060.5 3.434E+06 1.407E+06 0.410 3.208 4.5IIE+061.43 3.426E+06 1.403E+06 0.410 3.206 4.506E+062.86 3.425H06 1.402E+06 0.410 3.201 4.497E+06128

6.5.1.3. Edge-on geometry versus flat-on geometryFor the 25 keV monoenergetic beam source, the number of absorbed particles within thediamond was comparable in both the flat-on geometry and the edge-on geometry (Table 6.3).It was observed however, that the particle absorption fraction and energy deposited perparticle in diamond was about 7x higher in the edge-on than the flat-on geometry in the caseof the monoenergetic beam. This is because for the monoenergetic beam all impingingphotons have energies of 25 keV and are therefore generally more penetrating than in thespectrum case. There is thus a greater attenuation depth in edge-on geometry. Diamond inedge-on geometry profile showed barely twice as much absorption compared to the flat-ongeometry (Table 6.4) when the beam with an energy spectrum was incident on it but it wasfound to be 7x more absorbing (Table 6.3) with the monoenergetic beam source. As alreadystated this is attributed to the fact that the beam with an energy spectrum has a significantcomponent of lower energy photons, and since the edge-on and flat-on geometries havecomparable absorption levels (earlier geometry selection calculations) it is expected that bothgeometries would absorb the lower energy particles completely. The differences in the'fractional particles absorption' value and the 'deposited energy per particle' value observedfor the two geometries for photons with an energy spectrum incident on the crystal is aconsequence of the higher energy component of the spectrum where the flat-on geometry isfound to have lower absorption levels when compared to the edge-on geometry. The fact thatthe 'fractional particle absorption' value observed in edge-on is barely twice that ofthe flat-ongeometry, but that the corresponding 'deposited energy per particle' is about 2.6x larger thanthe flat-on geometry supports this conclusion.129

Table 6.3: Comparison of total deposited energies in diamond using edge-on and flat-on geometries for amonoenergetic, beam of 25 keV, as a function of beam aperture. The exposure areas of the edge-on and flat-onare 0.1 em' and I cm 2 respectively.Edge-on geometryBeam aperture Simulated Absorbedprimary Fractional particle Deposited energy Total depositedin degrees primary Particles absorption (keV) per primary energy (ke V)Particlesparticle0.01175 2.326£+06 5.923£+05 0.255 6.802 4.204£+060.175 2.427£+06 6.175£+05 0.254 6.797 4.197£+060.5 3.524£+06 6.169£+05 0.175 4.675 1.986£+061.43 9.384£+06 6.061£+05 0.065 1.724 2.702E+052.86 1.835£+07 5.879£+05 0032 0.856 6.656£+04Flat-on geometryBeam aperture Simulated Absorbedprimary Fractional particle Deposited energy Total depositedin degrees primary Particles absorption (keV) per primary energy (keV)Particlesparticle0.01175 1.424£+07 5.527£+05 0.039 1.036 5.847£+050.175 1.454E+07 5.640£+05 0.039 1.036 5.843E+050.5 1.454£+07 5.641£+05 0.039 1.036 5.844£+051.43 1.455E+07 5.642£+05 0.039 1.036 5.845£+052.86 1.463£+07 5.631£+05 0.038 1.028 5.753£+05130

Table 6.4: Comparison oftotal deposited energies in the edge-on and flat-on geometries using energy spectrum asa function ofbeam aperture. The exposure areas ofthe edge-on and flat-on are 0.1 em' and 1 em" respectively.Edge-on geometryBeam aperture Simulated Absorbed primary Fractional particle Deposited energy Total depositedin degrees primary Particles absorption (keV) per primary energy (ke V)Particlesparticle0.01175 2.082E+06 1.529E+06 0.735 8.312 1.330E+070.175 2. 178E+{)6 1.600£+06 0.735 8.307 1.329E+070.5 3.098E+{)6 1.602E+06 0.517 5.795 6.627£+061.43 8.242E+{)6 1.576E+06 0.191 2.141 8.917£+052.86 1.554E+{)7 1.487E+06 0.096 1.068 2.226E+05Flat-on geometryBeam aperture Simulated Absorbed primary Fractional particle Deposited energy Total depositedin degrees primary Particles absorption (ke V) per primary energy (keV)Particlesparticle0.01175 3.431E+06 1.406£+06 0.410 3.209 4.513E+060.175 3.432£+06 1.406E+06 0.410 3.209 4.512E+060.5 3.434E+06 1.407E+06 0.410 3.208 4.51IE+061.43 3.426E+06 1.403E+06 0.410 3.206 4.506E+062.86 3.425£+06 1.402£+06 0.410 3.201 4.497E+06131

6.6. Comparison ofdose values calculated from simulated results with those fromdirect measurementsThe number of particles impinging on the diamond 1\'d per cm 2 was calculated from the totalnumber ofparticles incident on the cone as:N dN, 0.1=- - , x--,Jr(r)- 0.785(6.3)where N; is the number of incident particles, r is the radius of the cone (half the diamondlength). The figures O. I and 0.785 represent the fraction of area covered by the diamond andthe entire area ofcone respectively. The total energy deposited Eted, was calculated as shown insection 5.1 (Interpretation ofsimulated results for the geometry selection study).The absorbed dose D in the diamond was calculated from the total energy deposited as:D = E"d 1m. (6.4)The units ofE ted , the total energy deposited should be in joules and m the mass of diamond inkg. For a diamond ofthickness 0./ em, area I cm 2 and density 3.515 g/cnr', ifthe total energydeposited has been calculated to be 1.33 x 10 7 keV, then the absorbed dose in the diamondwould be:D1.33 x 10' keV x 1.6 x 10- 16 J I keV- J0.3515 x 10-' kg-fGykg6.05 x l0-6Gy.(6.5)For a diamond ofvolume 0./ crn' (l x I x 0./ crrr') multiplying the total energy deposited inkeY by the conversion factor 4.549 x 10- 13 Gy.keyl gives the absorbed dose in Gy. From thecalculated dose and the number ofparticles incident on the diamond, the dose D p in Gy perphoton per cm 2 was calculated from:D = DP N, x 0.1(6.6)132

The results ofthe calculations have been compared with dose values from directmeasurements and presented in Table 6.5.Table 6.5: Comparison ofdose calculation from PENELOPE with those from directmeasurements at different kVps.Tube voltage, kVp . PENELOPE Direct Measurements(Nominal values) DoseabsorbedDirectMGD'Dose inequivalentMeasurements Thickness of PMMAin diamond(Gy/photons/crn")(Gy/photons/cm')(Gy/photons/cmi)25 1.73E-IO 2.05E-1O ± 0.54 E-IO 1.79E-IO ± 0,472 E-IO26 I.64E-1O 1.92E-1O ± 0.384 E-IO 1.67E-IO ± 0.334 E-IO28 1.47E-IO 1.62E-1O + 0.324 E-IO 1.4IE-IO + 0.282 E-IOThe comparable dose values expressed as dose per photon fluence in perspex and in diamondfor the three tube voltage settings, from direct measurements and simulation resultsrespectively stress the feasibility of obtaining total radiation dose values by using a diamondcrystal I em in depth as a sensing element. A variant of which is addressed in the followingsection.6.7. Alternative method for total radiation dose determination in Perspex(Subtraction method)In this section, a method to calculate total dose deposited in Perspex from the dose depositedin diamond is described. The Monte Carlo geometry file for the diamond probe in itscustomized Perspex phantom was set up to simulate the experimental conditions. The diamondsensor was, relative to the impinging radiation, in an "edge-on" geometry profile. Figure 6.3I MGD is mean glandular dose. Dose was calculated fora 20 rnm breastthickness.

PerspexphantomDiamondPerspexprobeziii i iImpingingradiationFigure 6.3: A 20 display of the geometry file of the diamond sensor, the Perspex probeand the Perspex phantom as viewed along the y-axis. Different colours were used forPerspex to contrast the different geometries.PerspexphantomDiamondziii i iImpingingradiationFigure 6.4: A 20 display of the geometry tile of the diamond sensor and the Perspexphantom as viewed along the y-axis.134

shows a 2D display of the geometry file of the diamond sensor, the Perspex probe and thePerspex phantom as viewed from along the y-axis with the impinging radiation incident alongthe z-axis. The energy deposited in each body (diamond sensor, Perspex probe and Perspexphantom) for different Perspex thicknesses were obtained using the PENELOPE code. ThePerspex probe has a circular shape for isotropic beam exposure. To account for the energy lossdue to the circular nature of the probe, the code was run for the diamond sensor using ageometry in which the Perspex is block or square shaped. The 2D display ofthe geometry fileof the diamond sensor in the block Perspex configuration is shown in Figure 6.4. The codewas also run for zero Perspex thickness in order to determine the deposited energy in diamondsensor with zero screening of the incident photons. Also obtained were energy depositedvalues for each Perspex thickness derived from the equation:(6.7)where E~ is the deposited energy in diamond without any Perspex, E;' is the deposited energyin diamond with Perspex of thickness Pt. Both E~and E;' are derived values from the code.E;' represents the calculated deposited energy value expected for Perspex ofthickness P" Theunderlying objective was to evaluate the exact difference of the energy deposited in diamondwith and without Perspex represents the energy deposited in Perspex for a non-monoenergeticbeam and when the diamond-sensing element is used in an "edge-on" profile. The simulationwas run for spectra energies of the kVps commonly used in X-ray mammographyexaminations and for Perspex phantom thicknesses of 15 to 100 rom. The results of thecalculations for 25,26,28 and 30kVp energy spectra are presented in Tables 6.6 to 6.13.135

The deposited energy in Perspex calculated usmg equation 6.7 was compared with thedeposited energy in Perspex as obtained directly from the code and their ratios are alsopresented. The ratio of the energy deposited in Perspex as given directly by the simulationoutput results and that calculated from the energies deposited in diamond, showed goodagreement. The difference in the ratios can be ascribed to the change in response of thediamond due to the change in spectral distribution between the 'after passing through thePerspex phantom' measurement and the direct 'in air measurement'. A factor (K) needs to beincorporated to account for the difference in response to 'beam hardening' caused by the twomaterials and be incorporated into the direct subtraction ofequation (6.7) as the formE P ' = K(E o - E P , )p d d (6.8)As the factor K reflects changes in beam profile (degree of beam hardening), the factor K isexpected to be different for different tube voltage settings, different X-ray tube targets andbeam filters. Although a larger value for higher kVp setting is implied, the simulated studysuggests that for Perspex, hence breast thicknesses greater than 30 mm, a constant K value canbe used for a fixed kVp and filter to derive the mean glandular dose value within ± 5%. For 25kVp X-ray energy spectrum, K is 0.95 (Table 6.6). The lower ratios of calculated depositedenergy in Perspex and the deposited energy in Perspex obtained directly from the code athigher tube voltage settings were due possibly to the presence of a larger proportion of highenergy components associated with higher tube voltage setting; not all of which are absorbedby the diamond. The general decrease in K values with decrease in Perspex thickness observedbelow 30 mm reflects a higher sensitivity of the method at these thicknesses, to the initialstages ofbeam hardening in which relatively high filtering ofthe low energy components fromthe impinging beam spectrum by the first few millimeters of the Perspex surface layers136

occurred, resulting in reduced number of photons impinging onto the diamond detector andcorresponding decrease in EX' hence increase in E;' values. From the simulated results itwould appear that for Perspex thickness less than 30 mm, employing the direct method ratherthan the subtraction method could lead to more accurate dose values.Table 6.6: Calculated deposited energy in Perspex probe of circular geometry using the energies deposited indiamond in "edge-on" geometry as a function of Perspex thickness. The energies deposited in Perspex as givenby the simulation output results are also shown. The results from a 2S kVp energy spectrum.Perspex Deposited energy Deposited energy Deposited energy % Ratio of simulatedThickness (keV) (keV) (keV) to calculated results(mm) per primary particle per primary particle per primary particlein Diamond in Perspex from in Perspex fromsimulation calculations' (alb) x 100(a)(b)0 14.045 0 0 010 2.985 9.803 11.060 8720 2.838 10.256 11.207 9130 0.833 12.343 13.212 9340 0.282 l3.074 l3.763 9550 0.098 13.361 13.947 9660 0.037 13.470 14.008 9670 0.013 13.520 14.032 9680 0.005 13.540 14.040 9690 0.002 l3.550 14.043 96.4100 0.002 13.551 14.044 96.4I b was calculated from deposited energies in diamond using the equation:137

Table 6.7: Calculated deposited energy in Perspex probe of block geometry using the energies deposited indiamond in "edge-on" geometry as a function of phantom thickness. The energies deposited in Perspex as givenby the simulation output results are also shown. The results from a 25 kVp energy spectrum.Perspex Deposited energy Deposited energy Deposited energy % Ratio of simulatedThickness (keV) (keY) (keV) to calculated results(nun) perprimaryparticle per primary particle per primary particlein Diamond in Perspex from in Perspexfromsimulation calculations (alb) x 100(a)(b)0 14.045 0 0 010 2.830 10.266 1l.215 9120 0.833 12.339 13.212 9330 0.282 13.070 13.763 9540 0.099 13.357 13.946 9650 0.037 13.469 14.008 9660 0.013 13.525 14.032 9670 0.006 13.542 14.039 9680 0.002 13.550 14.043 9690 0.002 13.551 14.043 96.4100 0.001 13.553 14.044 96.4138

Table 6.8: Calculated deposited energy in Perspex probe of circular geometry using the energies deposited indiamond in "edge-on" geometry as a function of Perspex thickness. The energies deposited in Perspex as givenby the simulation output results are also shown. The results from a 26 kVp energy spectrum.Perspex Deposited energy Depositedenergy Deposited energy % Ratio ofThickness (keV) (keV) (keV) simulated to(mm) per primary particle per primary particle per primary particle calculated resultsin Diamond in Perspex from inPerspex fromsimulationcalculations(a) (b) (a/b) x 1000 14.720 0 0 010 3.502 9.728 11.2\8 8520 3.366 10.226 11.354 8930 1.055 12.555 13.665 9140 0.367 13.455 14.353 9350 0.134 13.812 14.586 9460 0.051 13.951 14.669 9570 0.018 14.02 14.702 9580 0.008 14.05 14.712 9590 0.003 1406 14.717 95100 0.001 14.07 14.719 95139

Table 6.9: Calculated deposited energy in Perspex probe of block geometry using the energies deposited indiamond in "edge-on" geometry as a function ofphantom thickness. The energies deposited in Perspex as givenby the simulation output results are also shown. The results from a 26 kVp energy spectrum.Perspex Deposited energy Deposited energy Deposited energy % Ratio ofThickness (keV) (keV) (keV) simulated to(mm) per primary particle per primary particle per primary particle calculated resultsin Diamond in Perspex from in Perspex fromsimulation calculations(a) (b) (alb) x 1000 14.720 0 0 010 3.352 10.23 11.368 8920 1.055 12.56 13.665 9130 0.367 13.46 14.353 9340 0.134 13.81 14.586 9450 0.051 13.96 14.669 9560 0.018 14.02 14.702 9570 0.008 14.05 14.712 9580 0.003 14.06 14.717 9590 0.001 14.07 14.719 95100 0.001 14.07 14.719 95140

Table 6.10: Calculated deposited energy in Perspex probe of circular geometry using lbe energies deposited indiamond in "edge-on" geometry as a function of Perspex thickness. The energies deposited in Perspex as givenby the simulation output results are also shown. The results from a 28 kVp energy spectrum.Perspex Deposited energy Deposited energy Deposited energy % Ratio ofThickness (keV) (keV) (keV) simulatedto(mm) per primary particle per primary particle per primary particle calculated resultsin Diamond in Perspex from in Perspex fromsimulation calculations(a) (b) (alb) x 1000 15.989 0 0 010 4.574 9.51 11.415 8020 4.413 1O.D7 11.576 8530 1.566 12.85 14.423 8840 0.593 14.07 15.396 9150 0.236 14.65 15.753 92.560 0.095 14.90 15.894 9370 0.038 15.01 15.951 9480 0.016 15.06 15.973 9490 0.006 15.09 15.983 94100 0.003 15.10 15.986 94141

Table 6.11: Calculated deposited energy in Perspex probe of block geometry using the energies deposited indiamond in "edge-on" geometry as a function of phantom thickness. The energies deposited in Perspex as givenby the simulation output results are also shown. The results from a 28 kVp energy spectrum.Perspex Deposited energy Deposited energy Deposited energy % RatioofThickness (keV) (keV) (keV) simulated to(mm) per primary particle per primary particle per primary particle calculated resultsin Diamond in Perspex from in Perspex fromsimulationcalculations(a) (b) (alb) x 1000 15.989 0 0 010 4.395 10.08 11.594 8520 1.565 12.86 14.424 8830 0.594 14.07 15.396 9140 0.236 14.65 15.753 92.550 0.095 14.90 15.894 9360 0.038 15.01 15.951 9470 0.016 15.06 15.973 9480 0.006 15.09 15.983 9490 0.003 15.10 15.986 94100 0.001 15.10 15.988 94142

Table 6.12: Calculated deposited energy in Perspex probe of circular geometry using the energies deposited indiamond in "edge-on" geometry as a function of Perspex thickness. The energies deposited in Perspex as givenby the simulation output results are also shown. The results from a 30 kVp energy spectrum.Perspex Deposited energy Deposited energy Deposited energy o/u RatioofThickness (keV) (keV) (keV) simulated to(mm) per primary particle per primary particle per primary particle calculated resultsin Diamond in Perspex from in Perspex fromsimulationcalculations(a) (b) (alb) x 1000 17.24 0 0 010 5.660 9.238 11.58 7520 5.406 9.917 11.834 8130 2.083 13.063 15.157 8440 0.853 14.549 16.387 8750 0.355 15.314 16.885 9060 0.152 15.651 17.088 9170 0.064 15.841 17.176 9280 0.029 15.92 17.211 9290 0.015 15.96 17.225 92100 0.007 15.99 17.223 92143

Table 6.13: Calculated deposited energy in Perspex probe of block geometry using the energies deposited indiamond in "edge-on" geometry as a function of phantom thickness. The energies deposited in Perspex as givenby the simulation output results are also shown. The results from a 30 kVp energy spectrum.Perspex Deposited energy Deposited energy Deposited energy % Ratio ofThickness (keV) (keV) (keY) simulated to(mm) per primary particle per primary particle per primary particle calculated resultsin Diamond in Perspex from in Perspex fromsimulationcalculations(a) (b) (a/b) x 1000 17.24 0 0 010 5.394 9.924 11.846 8120 2.082 13068 15.158 8430 0.854 14.552 16.386 8740 0.355 15.315 16.885 8850 0.152 15.654 17.088 9160 0.064 15.839 17.176 9270 0.029 15.918 17.21I 9280 0.015 15.963 17.225 9290 0.007 15.986 17.223 92100 0.003 15.995 17.237 92144

Suggested furthermore is a method whereby the dose in Perspex (hence breast) ofthicknessesnormally encountered in mammography, can be determined from two measurements with thediamond probe using the subtraction method. An absorbed dose to glandular tissue of aparticular breast thickness can then be derived from the dose deposited in an equivalentPerspex thickness using either the Perspex to mean glandular dose conversion factors or theratio ofthe mass energy-absorption coefficient values ofglandular tissue or Perspex.6.8. Comparison of radiation dose in Perspex from experimental measurement withsimulated radiation dose in Perspex using PENELOPE MC codeRadiation dose to Perspex from mammography X-ray beam have been measured as describedin Chapter 5. The results ofthe measurements have been compared with the simulated dose toPerspex results and are presented in Figures 6.5 to 6.8.1.2" ~:50-l.l• Measured• Simulated25 kVpI r T T T T T1 11.0 T • • 1 I 1 • 10.90.8T~~I.2: " •ct 0.7"0

1.2I.l1.0" 0.900c 00..00~0.8">.~0;~0.70.6• Measured• Simulated26 kVpI•. II .I .II T T •I IT IT• 10.50.4 0 20 40 60 80 100Perspex thickness (rom)Figure 6.6: Comparison of measured radiation dose in Perspexwith radiation dose in Perspex simulated using PENELOPE code." 00c1.2• Measured1.\ • Simulated28 kVpT T T T T T• • • 1 1 1T • 11.0 T T • • ! I I !•0.9•1T •0.8 •00..00~ 1"-500.70;~ •0.60.50.4 0 20 40 60 80 100Perspex thickness (mm)figure 6.7: Comparison of measured radiation dose in Perspexwith radiation dose in Perspex simulated using PENELOPE code.146

~ "" 0 0-~"~" ;>.~1.2• Measured1.1 • Simulated30 kVp1.00.90.80; 0.7'"0.6•I .I .I .II TI• 1I I I0.50.4 0 20 40 60 80 100Perspex thickness (mm)Figure 6.8: Comparison of measured radiation dose in Perspexwith radiation dose in Perspex simulated using PENELOPE code.ConclusionsThe results of the exposure geometry selection study indicated that generally, the edge-onexposure geometry has higher absorption levels than the flat-on geometry and ought to be theX-ray exposure geometry of choice in radiation dose measurements especially in themammography X-ray energy range. The study has ShO\\l1 that using the diamond crystal in theedge-on geometry configuration, an absorption level of about 90% of the impinging photonscould be achieved. It has been established through the studies that both edge-on and flat-onexposure geometries have comparable absorption levels at lower X-ray energies. The flat-onexposure geometry is suggested to be appropriate for use in lower X-ray beam energies (lessthan 10 keV) as a higher response for the same crystal size can be expected.147

The exposure geometry selection study confirms that using a monoenergetic beam source,where lower energy components as contained in an X-ray energy spectrum are not included indose calculation, leads to dose misinterpretation. It is therefore imperative to use an energyspectrum to compute the deposited energy in any materia!. The contribution from the lowerenergy component of the spectrum must be accounted for since the photons with lower energycontribute significantly to the patient dose.Comparison of the results of absorbed dose in diamond crystal simulated using PENELOPEshowed good agreement with dose values from direct measurements. Similar comparisonshave been made for the simulated dose in Perspex with the measured dose in Perspex using thesubtraction method. For the later, the measured and simulated dose values showed goodagreement with each other.REFERENCESBar6 J., Sempau J, Femandez-Varea J.M., and Salvat F. (1994). Simplified MonteCarlo simulation ofelastic electron scattering in limited media. Nuc!. Instrum.Methods Phys. Res B 84,465-483.Bar6 J., Sempau J, Femandez-Varea J.M., and Salvat F. (1995). PENELOPE: An algorithmfor Monte carlo simulation of the penetration and energy loss of electrons andpositrons in matter. Nucl. lnstrum. Methods Phys. Res B 100, 31-46.Bielajew A.F., Hirayama H., Nelson W.R., and Rogers D.W.O. (1994). History, overviewand recent improvements of EGS4. National Research Council of Canada, ReportPIRS-0436.148

Briesmeister J.F. (1993). MNCP- A general Monte Carlo Nsparticle transport code LosAlamos National Laboratory Report LA-12625-M (Los Alamos, NM).Chetty I.J., Moran J.M., Nurushev T.S., McShan D.L., Fraass B.A., Wilderman SJ., andBielajew A.F. (2003). Experimental validation of the DPM Monte Carlo code usingminimally scattered electron beams in heterogeneous media. Phys. Med. BioI. 47,1837-1851.Dance D.R. (1980). The Monte Carlo calculation of integral radiation dose IIImammography. Phys. Med. BioI. 25, 25-37.Dance D.R. (1990). Monte Carlo calculation of conversion factors for the estimation ofmean glandular breast dose. Phys.Med. BioI. 35, 1211-1219.Doi K. and Heang-Ping C. (1980). Evaluation of absorbed dose in mammography: MonteCarlo simulation studies. Radiology 135, 199-208.Edmonds A.R. (1960). Angular momentum in quantum mechanics. 2 ndedition, PrincetonUniversity Press, Princeton.Halbleib J. (1989). Structure and Operation of the ITS code system in T. Jenkins, W.Nelson, A. Rindi, A. Nahum, and D. Rogers (Eds.). Monte Carlo Transport ofelectrons and photons. New York: Plenum Press, 249-262.Halbleib J.A., Kensek R.P., Mehlhorn T.A., Valdez G.D., Seltzer S.M., and Berger M.J.(1992). ITS version 3.0: The Integrated Tiger Series ofcoupled electron/photonMonte Carlo transport codes. Sandia Report SAND91-1634.Kramer R., Drexler G., Petoussi-Henss N., Zankl M., Regulla D., and Panzer W. (2001).Backscatter factors for mammography calculated with Monte Carlo methods.Phys. Med. BioI. 46, 771-781.149

Lalic D., Ilic R.D., and Stankovic S. J. (2001). Comparison ofmeasured and Monte Carlocalculated electron beam central axis depth dose in water. Archive ofOncology 9,83-87.Li J.S., Pawlicki T., Deng J., Jiang S.B., Mok E., and Ma C-M. (2000). Validation of aMonte Carlo dose calculation tool for radiotherapy treatment planning. Phys. Med.BioI. 45, 2969-2985.Llovet X., Sorbier L., Campos C.S., Acosta E. and Salvat F. (2003). Monte Carlo simulationof x-ray spectra generated by kilo-electron-volt electrons. J. Applied Phys. 93 (7),3844-3851.Nelson W.R., Hirayama H., and Rogers D.W.O. (1985). The EGS4 Code System ReportSLAC-265, Stanford Linear Accelerator Centor, Stanford, California.Rodriguez M.L. and deAlmeida C.E. (2002). Absorbed dose calculations III abrachytherapy pelvic phantom using the Monte Carlo method. J. Applied ClinicalMed. Phys. 3 (4),285-292.Salvat F., Fernandez-Varea J.M., Acosta E., and Sempau J. (2001). PENELOPE - A codesystem for Monte Carlo simulation of electron and photon transport, (OECDINEAData Bank, Issy-Ies Moulineaux, France, 2001). Available in PDF format fromwww.nea.fr.Salvat F., Femandez-Varea J.M., Bare J., and Sempau J. (1996). PENELOPE, an algorithmand computer code for Monte Carlo simulation of electron-photon showers. Ciemat(Centro de Investigaciones Energeticas, Medioambientales y Technologicas)Technical Report no. 799.150

Sempau J., Acosta E., Baro 1., Femandez-Varea J.M., and Salvat F. (1997). An algorithm forMonte Carlo simulation of coupled electron-photon transport. Nucl. Instrum. MethodsPhys. Res. B 132,377-390.Sempau J., Wilderman S.J., and Bielajew A. (2001). DPM, a fast, accurate Monte Carlo codeoptimized for photon and electron radiotherapy treatment plarming dose calculations.Department of Nuclear engineering and Radiological Sciences, The University ofMichigan, Ann Arbor, Michigan, USA.Siegbahn E.A., Nilson B., Femandez-Varea J.M., and Andreo P. (2003). Calculations ofelectron fluence correction factors using the Monte Carlo code PENELOPE. Phys.Med. BioI. 48,1263-1275.WaIling R.S. (PENEGRINE Program Office) (2000). PENEGRINE -3D Monte CarloMethod. Lawrence Livermore National Laboratory, Physics and AdvancedTechnologies Directorate, Livermore, CA 94550.Ye SJ., Brezovich LA., Pareek P., Popple R.A., and Naqvi S.A. (2004). Accuracy of lowenergyphoton cross sections for Monta Carlo dose calculations.Department ofRadiation Oncology, University of Alabama School of Medicine, Birmingham, AL35294, USA.lSI

Chapter 7GENERAL CONCLUSIONS152

Radiation doses from mammography X-ray beams have been assessed by various methods.The general conclusions from each ofthe methods employed are presented in this section.Mean glandular dose (MGD) has been estimated by the direct and spectral methods with goodagreement between the results from the two techniques. The study highlighted the parametersthat need to be taken into consideration in the calculation of MGD from both techniques inorder to achieve close agreement between results.For the direct method, determination of the X-ray beam quality; the tube loading (TL) forensuring the correct exposure ofthe standard phantom; measurement ofX-ray tube output forthe established tube loading and correction of the output values for the effect of ambienttemperature, pressure and humidity are important critical measurements. The results fromthese measurements affect the entrance surface air kerma (ESAK) directly and thereforerequire careful execution to avoid over or underestimation of ESAK and consequently, affectthe evaluated MGD values. In particular it has been observed that unless air density iscorrected for atmospheric conditions at the time of measurements, the estimated air kermacould have large errors especially for an X-ray spectrum with a high proportion oflow energyphoton contributions.The presence of scatter radiation onto the high purity germaruum detector during dataacquisition has been found to adversely influence the measurements and could lead toerroneously high ESAK and MGD values. The calculations of ESAK and MGD from spectraldata are based on large number oftheoretically generated attenuation coefficient data and thusaccurate data are needed to reduce errors and uncertainties in the measurements. To this end a153

fitting procedure for interpolating mass attenuation and mass energy-absorption coefficientsdata was developed. X-ray beam collimation and detector shielding needed to reduce radiationscattered onto the detector whilst acquiring spectral data; correction or stripping of spectraldata; accurate mass attenuation and mass energy-absorption coefficients for the energy regionof interest and modifying the X-ray beam quality (1 51 HVL) ofeach measured spectrum by theaddition of aluminium filters until it matches that from the ionisation chamber measurements,are the influencing factors that dictated the accuracy of MGD values calculated from spectralmeasurements.The study has shown that films with similar optical densities do not necessarily have similarcontrast values. It was also observed that optical density is not a direct reflection of thedeposited dose. The results ofthe study revealed that while the optical densities for some filmswere similar, their ESAK and corresponding MGD values were significantly different. It hasbeen found that while the use of a higher tube voltage on thinner phantom thickness results ina lower MGD, the effect on contrast is significantly detrimental and cannot be ignored. It isthus suggested that for the same X-ray target material and same filter type, lower tube voltagesshould be used to expose thinner thicknesses and higher tube voltages for larger thicknesses.The best range ofoptical density has been found to be 0.9 - 2.0 for thinner phantoms and 0.8­1.6 thicker phantoms. The equivalent contrast values are 0.25 - 0.38 and 0.2 - 0.34 for thinnerand thicker phantomsrespectively. Using optical densities outside these ranges couldcompromise contrast.154

Radiation dose from mammography X-ray beams has been evaluated using diamond as theactive radiation sensing material. To optimize its use as a radiation sensor, the geometry ofexposure of the diamond crystal was evaluated using theoretical calculations and also byPENELOPE Monte Carlo simulations. Highlighted from the geometry selection studies ofdeposited energy in a material, is the need to use an X-ray energy spectrum for accurateabsorbed dose interpretation, rather monoenergetic beam source. This ensures that thecontribution from the lower energy component of the spectrum is accounted for since thelower energy photons contribute significantly to the patient dose.The edge-on exposure geometry was found to have higher absorption levels than the flat-ongeometry and is suggested to be the X-ray exposure geometry of choice in radiation dosemeasurements especially in the mammography X-ray energy range. Using diamond as asensor in the edge-on geometry configuration, an absorption level of about 90% of theimpinging photons could be achieved.Two methods for dose measurements have been established in this study namely: the directmeasurement method and the subtraction method. Results ofabsorbed dose in diamond crystalsimulated using PENELOPE showed good agreement with dose values from directmeasurements. For the subtraction method, similar comparisons made for the simulated dosein Perspex with the measured dose in Perspex also showed good agreement with each other forPerspex (hence breast) ofthicknesses normally encountered in manunography at 25 to 30 kVpX-ray tube voltage settings.155

Using a feature of the specially designed diamond probe (Mark ITA) the important influence ofbreast compression plate in reducing radiation dose to patient has been systematically assessedin this study for different X-ray tube settings. It was observed that the 3 mm Perspex thicknesscommonly used as breast compression plate in mammography X-ray examination reducesradiation dose to the patient significantly. Reduction in radiation dose of 23%, 2I% and 16%were found for the 25, 26 and 28 kVp tube voltages respectively. In the case of a 30 kVp tubevoltage the radiation dose reduction was 14%. Further investigation into a reduction of dosethrough judicious choice of breast compression plate thickness for a particular tube voltagesetting whilst maintaining same image quality is suggested.The diamond probe that has been constructed out of this study has unique multiple featuresand thus is expected to find use for charged particles detection in the flat-on geometry as wellas possible application in electron and proton therapy. The diamond probe also has a furtherlikelihood ofbeing able to replace the different ionization probes, currently needed to providedose evaluation information; the exploitation of which awaits further evaluation. Fullcommercial exploitation of which however depends on the manufacturers being able toconsistently synthesize batches of crystal with such uniform and required performancecharacteristics as the samples used in the probe.From the study it is clear that crystals that require the often cited use of priming to improvetheir performances are not suited to be incorporated into probe for clinical use. In summary,this study systematically investigated many ofthe parameters that could affect the overall dosedeterminations and image quality and has outlined techniques of reducing or accommodating156

their influences. An alternative method of dose measurement that is less susceptible to errorsand can provide total radiation dose information is presented.157

Appendices158

Appendix 1APPLIED UNCERTAINTIES IN MEASUREMENTSThe percentage uncertainties and accuracy of measurements carried out in the course ofthis study are presented in this section. The uncertainty of a quantity such as tube outputis calculated based on the weighted values of possible sources of error associated witheach of the parameters involved in its measurement (European Protocol, 1996). Theuncertainties are calculated as the square root of the sum of squares of all parametersassociated with the quantity.I. Accuracy ofx-ray tube output measurementsi) Detector positional accuracy from actual point 0.1 ern error: ± 5%.ii) Diagnostic dosemeter accuracy and precision: ± I% and 05%.iii) Tube loading (mAs) meter accuracy: ± 5%.% Uncertainty =.J5' + I' + 05' + 5' = ± 7%2. Accuracy ofMGD from direct measurementsi) Detector positional accuracy 0.1 cm from actual point: ± 5%.ii) Accuracy and precision ofdosemeter: ± I% and ± 0.5%.iii) Accuracy and precision oftube loading (mAs) meter: both ± 5%.iv) Uncertainty due to phantom thickness: ± 5%.v) Uncertainty due to densitometer accuracy and precision: ± 0.02% and ±0.01% respectively.159

vi) Uncertainty due to ± 0.025 mm difference in HVL: ± 5%vii) Uncertainty in g (ESAK to MGD conversion factor): ± 10%viii) Uncertainty due to glandular tissue composition: ± 15%% Uncertainty =~52 + f +05 2 +5 2 +5 2 +0.02 2 +O.Oe+5 2 + 10 2 +15 2 = ± 20%3. Accuracy ofMGD from spectral measurementsi) Detector positional accuracy 0.1 em from actual point: ± 5%.ii) Accuracy and precision of tube loading (mAs) meter: both ± 5%.iii) Uncertainty due to ± 0.025 mrn difference in HVL: ± 5%iv) Uncertainty in g: ± 10%v) Uncertainty due to glandular tissue composition: ± 15%% Uncertainty = ~52 +5 2 +5 2 + 10 2 + 15 2 =± 20%4. Accuracy ofattenuation measurementsi) Detector positional accuracy from actual point 0.1 cm error: ± 5%.ii) Diagnostic dosemeter accuracy and precision: ± 1% and 05%.iii) Tube loading (mAs) meter accuracy: ± 5%.% Uncertainty = ~52 + e + 05 2 + 52 = ± 7%5. Accuracy HVL measurementsi) Diagnostic dosemeter accuracy and precision: ± 1% and 05%.160

ii) Tube loading (mAs) meter accuracy: ± 5%.% Uncertainty =.J5' + 1 2 + 0.5 2 = ± 5%6. Accuracy ofcontrast measurementsi) Densitometer accuracy: ± 0.02%ii) Phantom positional accuracy: ± 5%% Uncertainty =.J0.02 2 +5' =±5%161

Appendix 2LIST AND DESCRIPTION OF DIAMOND SAMPLESIn this section of the thesis, the diamond samples used for the study are listed inalphabetical order:Suite 2 No.2Dimension:Mass:Thickness 2 :N itrogen · 3 :Contact:Quality":Comments:Synthetic, single crystal, NiFe solvent/catalyst, AI gettering agent2.2 x2.6mm 29.37 ± 0.01 mg1.06 ± 0.001 mm16.9±2ppmSilver epoxy to boron p-doped and carbon-implanted surfaces.GoodBest detector, no space charge, needs no pre-irradiation.Name:Type:Dimension:Mass:Thickness:Nitrogen:ContactQuality:Comments:Sample A (20089A)Synthetic, polycrystaIIinelOx lOmm 2173.8 ±O.OOI mg0.52 ± 0.001 mm50 ± lOppbTitanium to platinum and gold in the surface.GoodSpace charge effect needs no pre-irradiation.'Type indicates diamondtype as well as substrate andcatalyst used in the production.2 Thicknessrefersto the distancebetween the polished surface andthe contactsurface.3 Nitrogen concentration present in sample.4 Qualityrefers to % non-diamond or graphitic carbon, and % 13c.162

Name:Type:Dimension:Mass:Thickness:Nitrogen:Contact:Quality:Comments:Sample B (20089B) No.1Synthetic, polycrystalline10 x 10 mm"323.84 ± 0.001 mg1.02 ± 0.001 mm10 ±2 ppbTitanium to platinum and gold in the surface.UnsatisfactorySpace charge effects, needs pre-irradiation.Name:Type:Dimension:Mass:Thickness:Nitrogen:Contact:Quality:Comments:Sample B (20089B) No.2Synthetic CVD polycrystalline diamondlOx IOmm 2330.88 ± 0.001 mg1.02 ± 0.001 mm10 ±2 ppbTitanium to platinum and gold in the surface.UnsatisfactorySpace charge effects, needs pre-irradiation.Name:Type:Dimension:Mass:Thickness:Nitrogen:Contact:Quality:Comments:Sample FSynthetic HPHT single crystal diamond8.10 x 7.52 mm'178±0.00Img1.38 ± 0.001 mm130±2 ppmTitanium to platinum and gold in the surface.GoodNo space charge effect, needs no pre-irradiation.163

Appendix 3PUBLICAnONS, PATENTS AND CONFERENCESPublicationsAssiamah M., Mavunda RD., Nam T.L., and Keddy R.J. (2003). Segmented multifit ofpolynomial function for mass attenuation and energy-absorption coefficientvalues. Radiat. Phys. Chern. 67 (1), 1-6.Assiamah M., Mavunda RD., Nam T.L., and Keddy RJ. (2003). Effect ofpressure,temperature and humidity in airon photon fluence and air kerma values at low photonenergies. Radial. Phys. Chern. 68 (5), 707-720.Assiamah M., Nam T.L., and Keddy R.J. (2004). Dosimetric techniques formammography X-ray beams. Radial. Phys. Chern. 71,957-958.Mavunda R.D., Assiamah M., Nam T.L., and Keddy RJ. (2004). Bremsstrahlung spectrafrom diagnostic X-rays. Radial. Phys. Chern. 71, 991-992.Assiamah M., Nam T.L., and Keddy RJ. Comparison of mammography radiation dosevalues obtained from direct incident air kerma measurements with values frommeasured X-ray spectral data. Accepted for publication in Applied Radiation andIsotopes.PatentsNam T.L., Assiamah M., and Keddy RJ. (2004). Improvement in Radiation Detectors. SAProvisional Patent 2004/7675.ConferencesAssiamah M., Mavunda RD., Narn T.L., and Keddy RJ. Comparison ofmammography dosemeasurements with dose values calculated from computer generated X-ray spectrum.164

42"" Congress and Winter School ofthe South African Association of Physicists inMedicine and Biology, 29 July to 2 August 2002, Pretoria, South Africa.Mavunda RD., Assiamah M., NamTL, and Keddy R.J. X-ray spectra measurements withinthe mammography energy range for dose calculations. 42 nd Congress and WinterSchool ofthe South African Association ofPhysicists in Medicine and Biology, 29July to 2 August 2002, Pretoria, South Africa.'Assiamah M., Nam TL., and Keddy RJ. Dosimetric techniques for mammography x-raybeams. 9'h International Symposium ofRadiation Physics (ISRP-9) and Workshop onRadiation Based Analytical Techniques, 24 - 31 October 2003, Cape Town, SouthAfrica.Mavunda RD., Assiamah M., NamTL, and Keddy RJ. Bremsstrahlung spectra fromdiagnostic X-rays. 9'h International Symposium ofRadiation Physics (ISRP-9) andWorkshop on Radiation Based Analytical Techniques, 24 - 31 October 2003, CapeTown, South Africa.Assiamah M., Nam TL, and Keddy RJ. CVD diamond as a radiation sensing element formammography X-ray beam dose measurements. 44'h Congress and Winter School ofthe South African Association of Physicists in Medicine and Biology, 20 - 23September 2004, Johannesburg, South Africa.Assiamah M., Nam TL., and Keddy R.J. Comparative studies ofmeasured mammographyX-ray beam doses with theoretical calculations from the Monte Carlo codePENELOPE using CVf) diamond as radiation sensor. 44 th Congress and WinterSchool of the South African Association of Physicists in Medicine and Biology,20 - 23 September 2004, Johannesburg, South Africa .• Awardedprize for "Best Poster', 9 t h. International Symposiumof Radiation Physics (ISRP-9) andWorkshop on Radiation Based Analytical Techniques, 24 - 31 October 2003, CapeTown, South Africa.165