ROLE OF ADVANCED 2 AND 3-DIMENSIONAL ULTRASOUND FOR DETECTING PROSTATE CANCER K. C. BALAJI, WILLIAM R. FAIR, ERNEST J. FELEPPA, CHRISTOPHER R. PORTER, HAROLD TSAI, TIAN LIU, ANDREW KALISZ, STELLA URBAN AND JOHN GILLESPIE From the University of Nebraska Medical Center, Omaha, Nebraska, Haelth, Riverside Research Institute and New York Presbyterian Medical Center, New York and State University of New York Medical Center, Stony Brook, New York, New York, Kaiser-Permanente, Los Angeles, California, and National Cancer Institute, Bethesda, Maryland ABSTRACT Purpose: We explored the clinical usefulness of spectrum analysis and neural networks for classifying prostate tissue and identifying prostate cancer in patients undergoing transrectal ultrasound for diagnostic or therapeutic reasons. Materials and Methods: Data on a cohort of 215 patients who underwent transrectal ultra- sound guided prostate biopsies at Memorial-Sloan Kettering Cancer Center, New York, New York were included in this study. Radio frequency data necessary for 2 and 3-dimensional (D) computer reconstruction of the prostate were digitally recorded at transrectal ultrasound and prostate biopsy. The data were spectrally processed and 2-D tissue typing images were generated based on a pre-trained neural network classification. We used manually masked 2-D tissue images as building blocks for generating 3-D tissue images and the images were tissue type color coded using custom software. Radio frequency data on the study cohort were analyzed for cancer probability using the data set pre-trained by neural network methods and compared with conventional B-mode imaging. ROC curves were generated for the 2 methods using biopsy results as the gold standard. Results: The mean area under the ROC curve plus or minus SEM for detecting prostate cancer for the conventional B-mode and neural network methods was 0.66  0.03 and 0.80  0.05, respectively. Sensitivity and specificity for detecting prostate cancer by the neural network method were significantly increased compared with conventional B-mode imaging. In addition, the 2 and 3-D prostate images provided excellent visual identification of areas with a higher likelihood of cancer. Conclusions: Spectrum analysis could significantly improve the detection and evaluation of prostate cancer. Routine real-time application of spectrum analysis may significantly decrease the number of false-negative biopsies and improve the detection of prostate cancer at transrectal ultrasound guided prostate biopsy. It may also provide improved identification of prostate cancer foci during therapeutic intervention, such as brachytherapy, external beam radiotherapy or cryotherapy. In addition, 2 and 3-D images with prostate cancer foci specifically identified can help surgical planning and may in the distant future be an additional reliable noninvasive method of selecting patients for prostate biopsy. KEY WORDS: prostate; prostatic neoplasms; imaging, three-dimensional; ultrasonography; spectrum analysis Transrectal ultrasound guided prostatic biopsy is currently the standard for detecting prostate cancer in the United States. In a study of 132,246 men undergoing prostate biopsy for the first time 38.2% were diagnosed with prostate cancer. Repeat biopsy within a year in 6,350 of the remaining men revealed prostate cancer in 25.2%, suggesting a significant false-negative rate of more than 40% at initial biopsy. 1 The reported sensitivity, specificity and positive predictive values of hypoechoic lesions of the prostate on transrectal ultra- sound, which are considered characteristic of prostate cancer, are only 85.5%, 28.4% and 29%, respectively, for diagnosing prostate cancer. 2 Recently considerable advances have been made in medical ultrasonography, including transducer tech- nology, color Doppler imaging, echo contrast, ultrasonic tem- perature sensing and high intensity focused ultrasound. 3, 4 Furthermore, using echo contrast with power Doppler tech- nology in patients with suspected prostate cancer improves the sensitivity of detecting prostate cancer by about 50%, while maintaining 80% specificity. 5 In the last 2 decades computational technology has been increasingly used in medicine and neural networks are cur- rently being investigated for clinical use in urology. 6 Because conventional B-mode ultrasound has limited capability for detecting prostate cancer, we investigated the combined use- fulness of spectrum analysis of radio frequency data acquired during transrectal ultrasound of the prostate and neural network methods for improving prostate cancer detection. We investigated whether classification based on spectrum analysis parameters by neural network methods provide bet- ter sensitivity and specificity than conventional B-mode im- aging. In addition, we constructed 2 and 3-dimensional (D) images of the prostate using color codes to indicate cancer likelihood, which in the future may prove invaluable for real-time imaging of prostate cancer during transrectal ul- trasound for diagnostic or therapeutic purposes. MATERIALS AND METHODS Transrectal ultrasound guided prostate biopsy. The cohort of 215 patients in this study underwent biopsy for abnormal pros- Accepted for publication July 5, 2002. Supported by National Institutes of Health Grant CA53561. 0022-5347/02/1686-2422/0 Vol. 168, 2422–2425, December 2002 THE JOURNAL OF UROLOGY ® Printed in U.S.A. Copyright © 2002 by AMERICAN UROLOGICAL ASSOCIATION,INC. ® DOI: 10.1097/01.ju.0000036435.13421.57 2422