Official Proceedings - AIUM

More documents

Recommendations

Info

American Institute of Ultrasound in Medicine <strong>Proceedings</strong> J Ultrasound Med 32(suppl):S1–S134, 20131538495 Evaluation of Cervical Cancer Detection With AcousticRadiation Force Impulse Ultrasound Imaging: PreliminaryResultsYijin Su,* Lianfang Du, Ying Wu, Juan Zhang, XuemeiZhang, Xiao Jia, Yingyu Cai, Yunhua Li, Jing Zhao, QianLiu School of Medicine, Shanghai Jiaotong University,Shanghai, ChinaObjectives—To evaluate the application of acoustic radiationforce impulse (ARFI) ultrasound imaging and its potential value for characterizingcervical cancer.Methods—ARFI of the uterine cervix was performed in 58 patientswith cervical cancer before operation. Elastographic imaging (EI),Virtual Touch tissue imaging (VTI), and Virtual Touch tissue quantification(VTQ; Siemens Medical Solutions, Mountain View, CA) wereused to qualitatively and quantitatively analyze the elasticity and hardnessof lesions.Results—Compared to the surrounding cervix tissue, the EI,VTI, and VTQ images between malignant lesions and surrounding normaltissues showed a significant difference (P < .001).Conclusions—ARFI of the uterine cervix may be an objectivemethod for assessment of softening of tissue. It has high sensitivity andspecificity in evaluating cervical cancer and therefore has good diagnosticvalue in clinical applications.Table 1. Comparison of VTQ Between Cervical Cancer and Normal TissueTissue Cases, n Mean, m/s SD, m/sCervical cancer 58 3.41 1.59Normal tissue 58 2.12 1.27Compared with normal tissue, VTQ of cervical cancer was higher (P < .001).1512000 Texture-Based Ovarian Tumor Characterization Using3-Dimensional UltrasoundU. Rajendra Acharya, 1 Stefano Guerriero, 2 Filippo Molinari, 3Luca Saba, 4 Jasjit Suri 5,6 *1 Electronics and Computer Engineering,Ngee Ann Polytechnic, Singapore; 2 Obstetrics andGynecology, University of Cagliari, Cagliari, Italy; 3 Electronicsand Telecommunications, Politecnico Torino, Torino, Italy;4Radiology, Azienda Ospedaliero Universitaria di Cagliari,Cagliari, Italy; 5 Global Biomedical Technologies, Roseville,California USA; 6 Biomedical Engineering, Idaho State University,Pocatello, Idaho USAObjectives—Among gynecologic malignancies, ovarian canceris the most frequent cause of death. Differential diagnosis is difficult,thus exposing patients to unneeded surgical treatment. We developed acomputer-aided diagnostic technique that uses ultrasound images of theovary to accurately classify benign and malignant ovarian tumors.Methods—Twenty women (age range, 29–74 years; mean ±SD, 49.5 ± 13.48 years), 11 premenopausal and 9 postmenopausal, wererecruited for this study. The histologic specimens revealed 10 malignantand 10 benign lesions. Prior to surgery, each patient was associated witha 3D volume of 100 images. Feature extraction was made by using localbinary pattern and laws texture energy. The data were used to train a classifierbased on a support vector machine (SVM) with 5 different kernels.The data set was randomly split into 10 equal folds, each fold containingthe same ratio of nonrepetitive samples from both classes (malignant andbenign). At each iteration, 9 folds were used to train the SVM, and 1 foldwas classified. We iterated the procedure 10 times to explore all the possiblecombinations. The averages of the performance metrics obtained inall the iterations are reported as the overall performance metrics.Results—The performance metrics obtained on training theSVM classifier of various kernel configurations using the 14 significantfeatures are reported in Table 1. All the kernels demonstrated excellentability in classifying the samples from both classes. The highest accuracyof 99.9% was registered by the radial basis function (RBF) kernel.Conclusions—The novelty of this study is the use of low-costultrasound images and a highly discriminating combination of simple texturefeatures fed to an SVM classifier to obtain the highest accuracy ofnearly 100% in ovarian tumor classification.Table 1. Classifier PerformanceSVM Accuracy, Sensitivity, Specificity, PPV,Kernel TP TN FP FN % % % %Linear 100 99 0 1 99.8 99.6 100 100Poly 1 100 100 0 0 99.8 99.6 100 100Poly 2 100 100 0 0 99.9 100 99.9 99.9Poly 3 100 100 0 0 99.8 99.9 99.8 99.8RBF 100 100 0 0 99.9 100 99.8 99.8FN indicates false-negative; FP, false-positive; PPV, positive predictive value;TN, true-negative; and TP, true-positive.1512001 Tumor Characterization From 3-Dimensional GynecologicUltrasound: A New Online Feature-Based ParadigmU. Rajendra Acharya, 1 Luca Saba, 2 Filippo Molinari, 3Stefano Guerriero, 4 Jasjit Suri 5,6 * 1Electronics and ComputerEngineering, Ngee Ann Polytechnic, Singapore; 2 Radiology,Azienda Ospedaliero Universitaria di Cagliari, Cagliari,Italy; 3 Electronics and Telecommunications, Politecnico Torino,Torino, Italy; 4 Obstetrics and Gynecology, University ofCagliari, Cagliari, Italy; 5 Global Biomedical Technologies,Roseville, California USA; 6 Biomedical Engineering, IdahoState University, Pocatello, Idaho USAObjectives—Among gynecologic malignancies, ovarian canceris the most frequent cause of death. Differential diagnosis is difficult, exposingpatients to unneeded surgical treatment. The objective of this workwas to develop a computer-aided diagnostic (CAD) technique that uses 3Dacquired ultrasound images of the ovary and image-mining algorithms tocharacterize and classify benign and malignant ovarian tumors.Methods—Twenty women (age range, 29–74 years; mean ±SD, 49.5 ± 13.5 years), 11 premenopausal and 9 postmenopausal, were recruitedfor this study. The histologic specimens revealed 10 had malignantand 10 had benign lesions. Prior to surgery, each patient wasassociated with a 3D volume of 100 images. We extracted features basedon the textural changes in the image and also features based on the higherorderspectra (HOS) information. The significant features were then selectedand used to train and evaluate the decision tree (DT) classifier.The data set was randomly split into 10 equal folds, each fold containingthe same ratio of nonrepetitive samples from both the classes (malignantand benign). At each iteration, 9 folds were used to train the DT, and 1fold was classified. We iterated the procedure 10 times to explore all thepossible combinations. The averages of the performance obtained in all theiterations are reported as the overall performance.Results—The simple DT classifier presented high accuracy of95.1%, sensitivity of 92.5%, and specificity of 97.7%. Full performanceis given in Table 1.Conclusions—A novel combination of 4 texture and HOSbased features that adequately quantify the nonlinear changes in both benignand malignant ovarian ultrasound images was used to develop classifiers.The CAD tool would be a more objective alternative to manualanalysis of ultrasound images, which might result in interobserver variations.The system can be installed as a stand-alone software application inthe physician’s office at no extra cost.S47
American Institute of Ultrasound in Medicine <strong>Proceedings</strong> J Ultrasound Med 32(suppl):S1–S134, 2013Table 1. Classifier PerformanceAccuracy, Sensitivity, Specificity, PPV,TN FN TP FP % % % %DT 98 8 93 2 95.1 97.8 92.5 97.7FN indicates false-negative; FP, false-positive; PPV, positive predictive value; TN,true-negative; and TP, true-positive.1537972 Value of Transvaginal Ultrasonography in DiagnosingAdenomyosisBirsen Ogutcu, 1,2 * Kathryn Gunnison, 2 Bakytbubu Arynova, 1Jack Garon, 1,2 Josef Blankstein 1,2 1 Obstetrics and Gynecology,Mount Sinai Hospital, Chicago, Illinois USA; 2 Obstetricsand Gynecology, Rosalind Franklin University of Medicine andScience, North Chicago, Illinois USAObjectives—The purpose of this study was to look at ultrasonographicfindings of uteri with adenomyosis confirmed by histopathologyafter hysterectomy.Methods—The ultrasound reports and images from 76 patientswho underwent hysterectomy with a diagnosis of adenomyosis on pathologicreports were evaluated. Data collection included the myometrialechogenicity (focal or diffuse heterogeneous echo texture vs homogeneousecho texture), the presence or absence of myometrial cysts, the presenceor absence of subendometrial linear striations, and the quality of the endometrial/myometrialborder (ill defined vs well defined).Results—Of the 73 ultrasound images reviewed, 70 (96%)were shown to have a heterogeneous myometrial echo texture. Of these 70,11 (26%) were described as focal heterogeneity. Myometrial cysts werefound in 53 (72%) of the images. Forty-six (63%) images were shown tohave an ill-defined endometrial/myometrial border. Thirty-two (44%) imageswere shown to have subendometrial linear striations. Of the imagesshown to have myometrial cysts, all 53 (100%) were shown to have a heterogeneousmyometrial echo texture.Conclusions—In conclusion, based on our retrospective studiesand others, transvaginal ultrasonography is a cost-effective, noninvasive,and widely available method to look for ultrasonographic findingsthat are common in uteri with pathologically confirmed adenomyosis. Inthis study, we identified a heterogeneous myometrial echo texture as themost common ultrasonographic finding in uteri with confirmed adenomyosis.Additional prospective studies to further determine the accuracyand reliability of transvaginal ultrasonography as a tool to diagnose adenomyosisare needed.1509704 To Chaperone or Not to ChaperoneJenny Parkes, 1,2 * Michal Schneider-Kolsky, 2 Paul Lombardo 21Australian School of Medical Imaging, Sydney, New SouthWales, Australia; 2 Medical Imaging and Radiation Science,Monash University, Melbourne, Victoria, AustraliaObjectives—Chaperone use during transvaginal (TV) ultrasoundis variable. This paper discusses such use in regard to both the sonographer/sonologistand patient’s perspectives.Methods—A survey of sonographer members of the AustralianSonographers Association was performed online to audit current practice;in addition, a literature review of current international practices, protocols,and opinions was undertaken.Results—Three-hundred fifty of 2219 (15.8%) sonographersparticipated in this survey. Most sonographers surveyed used a chaperoneoccasionally (42.9%), with 70% preferring to never use a chaperone.Patient embarrassment and lack of privacy were key issues, findings thatwere supported by the literature. Female sonographers used chaperonesoccasionally or never (89%), while 60% of male sonographers used achaperone always or most of the time to comply with departmental protocolsand for medicolegal protection. Sonographers in private specialistobstetric and gynecologic centers where TV pelvic ultrasound is commonlya routine part of the examination never, or only occasionally, useda chaperone. Review of the current literature and protocols highlightedthat the use of a chaperone should consider the patient’s privacy, embarrassment,and preference. In regard to the sonographer/sonologist, selectionof an appropriate chaperone, availability, and the informed consent ofthe patient are important considerations.Conclusions—Chaperones for TV scanning are predominantlyused by male sonographers in Australia. It is unclear if the patient preferencefor a chaperone to be present is taken into account. The appropriateprotection for both sonographers/sonologists and patients needs to be considered,in particular, privacy issues. Protocols should be developed totake all of these factors into account.1538726 What Causes Postmenopausal Bleeding (PMB)? A ProspectiveStudy of 670 Consecutive Patients With PMB ExaminedWith Regard to Body Mass Index and the Prevalenceof Abnormalities on SonohysterographyAlex Hartman,* Rose Lee, Brian Hartman Imaging, TrueNorth Imaging, Thornhill, Ontario, CanadaObjectives—The causes of postmenopausal bleeding (PMB)are myriad. This prospective study examines the causes of PMB and correlatesthese findings to patient body mass index (BMI).Methods—Sonohysterography was performed on 1108 consecutivepostmenopausal women at an academically oriented private practicefrom October 2010 to August 2011. Six hundred seventy of thepatients had PMB. The prevalence of uterine abnormalities and the patient’sBMI were obtained for each group. χ 2 tests of independence and1-way analysis of variance were performed to determine significant differences.Results—Of the 670 with PMB, 262 (39.1%) had endometrialpolyps; 273 (40.1%) had fibroids; 39 (6.8%) had submucosal fibroids; and369 (55%) had a normal cavity, with no polyps, or submucosal fibroids.There was a statistically significant difference between the mean BMI ofpatients with endometrial polyps (28.34) and those with normal cavities(27.33) and only fibroids (27.64) (F = 2.95; df = 3; P = .03).Conclusions—More than half of patients with PMB have anormal uterine cavity, and almost 40% have endometrial polyps. Patientswith endometrial polyps have a higher BMI than those without polyps.Table 1Patients, n Patients, % BMINo uterine abnormality 233 34.8 27.33Intramural fibroid 136 20.3 27.64Polyp 164 24.5 28.34Submucosal fibroid 39 5.8 27.60Polyp + intramural fibroid 91 13.6 28.80Polyp + submucosal fibroid 7 1.0 28.76Total 670 100 27.931541040 Spatial Variability of Shear Wave Speed Estimation in theNormal Nonpregnant CervixLindsey Carlson, 1 * Mark Palmeri, 2 Lisa Reusch, 1 HelenFeltovich, 1,3 Timothy Hall 1 1 Medical Physics, University ofWisconsin, Madison , Wisconsin USA; 2 Biomedical Engineering,Duke University Pratt School of Engineering, Durham,North Carolina USA; 3 Maternal-Fetal Medicine, IntermountainHealthcare, Park City, Utah USAObjectives—Throughout pregnancy, beginning soon after conception,the cervix remodels, resulting in softening, shortening, and dilationto allow for eventual delivery of a fetus. Premature remodeling mayS48
Page 1 and 2:
2013Journal ofUltrasoundin Medicine
Page 5 and 6:
Small-Animal Preclinical High-Frequ
Page 7 and 8:
2013 AIUM Award WinnersJoseph H. Ho
Page 9 and 10:
2013 AIUM Award WinnersDistinguishe
Page 11 and 12:
Endowment for Education andResearch
Page 13 and 14:
Rosa De Vermette, MD, RDCS,RDMS, RO
Page 15 and 16: Kee-Hak Lim, MDG. Sharat Lin, PhDTo
Page 17 and 18: David Stark, MDA. Thomas Stavros, M
Page 19 and 20: *Presenter of scientific paper with
Page 21 and 22: American Institute of Ultrasound in
Page 65: American Institute of Ultrasound in
Page 116 and 117:
American Institute of Ultrasound in
Page 118 and 119:
Page 120 and 121:
Page 122 and 123:
Page 124 and 125:
Page 126 and 127:
Page 128 and 129:
Page 130 and 131:
Page 132 and 133:
Page 134 and 135:
Page 136 and 137:
Page 138 and 139:
Page 140 and 141:
Page 142 and 143:
Page 144 and 145:
Page 146 and 147:
Page 148 and 149:
Page 150 and 151:
Page 152 and 153:
Page 154:
© Greater Phoenix CVB
show all

Official Proceedings - AIUM

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?