Bush__The_Essential_Physics_for_Medical_Imaging - Biomedical ...

511-keV collimators and because the camera heads must contain adequate shieldingfor 511-ke V photons from outside the field of view.Double- and triple-head SPECT cameras with coincidence detection capabilityare commercially available. These cameras have planar NaI(TI) crystals coupledto PMTs. With standard nuclear medicine collimators mounted, they function asstandard scintillation cameras and can perform planar nuclear medicine imagingand SPECT. With the collimators removed or replaced by axial collimators, theycan perform coincidence imaging of positron-emitting radiopharmaceuticals.When used for coincidence imaging, these cameras provide spatial resolutionsimilar to that of a dedicated PET system. However, in comparison to a dedicatedPET system, they detect a much lower number of true coincidences. The smallerattenuation coefficient of sodium iodide and the limited crystal thickness limits theintrinsic efficiency of each detector. For example, the photopeak efficiency of a 0.95cm ('Is inch) thick NaI(TI) crystal for annihilation photons is about 12% (see Chapter21, Fig. 21-9). The fraction of annihilation photon pairs incident on a pair ofdetectors that yields true coincidences is equal to the square of the photopeak efficiency(assuming that only photopeak-photopeak coincidences are registered).Thus, for 0.95-cm thick crystals, only about 1.5% of annihilation photon pairsreaching the detectors are registered as true coincidences. For this reason, manysuch cameras have thicker NaI crystals than standard scintillation cameras, withthicknesses ranging from 1.27 to 2.5 cm ('/2 to 1 inch). Even modest increases incrystal thickness significantly improve the number of true coincidences detected.However, even with thicker crystals, the coincidence detection sensitivities of multiheadscintillation cameras with coincidence circuitry are much lower than those ofdedicated PET systems. The lower coincidence detection sensitivity causes morestatistical noise in the images than in images from a dedicated PET system. Futthermore,the low coincidence detection sensitivity forces the use of measures, suchas wide spacing of axial septa in two-dimensional acquisition or three-dimensionalacquisition and wide energy discrimination windows, to obtain adequate numbersof true coincidences. These measures increase the scatter coincidence fraction, producingimages of lower contrast than a dedicated PET system. Both the greaternoise due to fewer coincidences and the reduced contrast due to the higher scattercoincidence fraction limit the ability to detect small lesions in the patient, in comparisonto a dedicated PET system. When these cameras are used with collimatorsas standard scintillation cameras, their thicker crystals produce slightly worse intrinsicspatial resolution for lower energy photons, such as those emitted by T c 99mand TI 201, but have little effect on the system spatial resolution.Some manufacturers provide two-dimensional data acquisition, whereas othersprovide three-dimensional acquisition (described earlier in this chapter). The camerasusing two-dimensional data acquisition are provided with axial collimators toreduce the singles rates from out-of-slice activity and to reduce the fraction of scattercoincidences.Because these cameras are either uncollimated or used with axial collimatorswhen performing coincidence imaging, they are subject to very high interactionrates and must have very fast electronics. Because such a system has only two orthree detectors (scintillation camera heads) instead of the hundreds of detectors ina dedicated PET system, its performance suffers more from dead-time count losses.Although many interactions from lower energy photons are rejected by theenergy discrimination circuits of the camera, they still cause dead-time count lossesand can interfere with the determination of the energies and positions of interactions.