Mp31-18 Artificial Intelligence-Assisted Prostate Cancer Detection on B-Mode Transrectal Ultrasound Images

You have accessJournal of UrologyProstate Cancer: Detection & Screening III (MP31)1 May 2024MP31-18 ARTIFICIAL INTELLIGENCE-ASSISTED PROSTATE CANCER DETECTION ON B-MODE TRANSRECTAL ULTRASOUND IMAGES Indrani Bhattacharya, Sulaiman Vesal, Hassan Jahanandish, Moonhyung Choi, Steve Zhou, Zachary Kornberg, Elijah Richard Sommer, Richard E. Fan, James D. Brooks, Mirabela Rusu, and Geoffrey A. Sonn Indrani BhattacharyaIndrani Bhattacharya , Sulaiman VesalSulaiman Vesal , Hassan JahanandishHassan Jahanandish , Moonhyung ChoiMoonhyung Choi , Steve ZhouSteve Zhou , Zachary KornbergZachary Kornberg , Elijah Richard SommerElijah Richard Sommer , Richard E. FanRichard E. Fan , James D. BrooksJames D. Brooks , Mirabela RusuMirabela Rusu , and Geoffrey A. SonnGeoffrey A. Sonn View All Author Informationhttps://doi.org/10.1097/01.JU.0001008936.35187.0b.18AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Prostate cancer detection on b-mode transrectal ultrasound (TRUS) images is challenging due to imaging artifacts, low signal-to-noise ratio, and benign confounders, with cancer detection rates around 40%. While adding MRI-guided targeted biopsy improves clinically significant prostate cancer (csPCa) detection, ∼80% of biopsies in the United States are performed without pre-biopsy MRI due to high cost and limited access. Thus, improving csPCa detection and localization on TRUS would represent a major clinical advance. METHODS: We developed an Artificial Intelligence (AI) system that detects csPCa on b-mode TRUS images. Our system first learns ultrasound biomarkers emphasizing cancer regions through registration-independent multimodal image correlations with MRI and whole-mount histopathology images. These learned ultrasound biomarkers are then used as inputs to a convolutional neural network model to detect csPCa. MRI and histopathology images are only required during model training to identify MRI-informed and histopathology-informed ultrasound biomarkers, and are not required when applying the model to future cases where only TRUS is available. We trained the system using an internal cohort of 289 men who underwent MRI-TRUS fusion biopsies and/or radical prostatectomy, and an additional publicly available external cohort of 1573 men with MRI-TRUS fusion biopsies. We evaluated the system on a lesion level and a patient level using 41 internal patients who underwent radical prostatectomy with whole-mount digital histopathology registered to MRI and TRUS to create accurate gold standard labels. We also performed a multi-reader study where we compared AI performance with four urologists who reviewed TRUS images while blinded to clinical information but provided with unlimited time for image review. RESULTS: Our AI system achieved patient-level sensitivity and specificity of 0.65 and 0.81, and lesion-level sensitivity and specificity of 0.60 and 0.80 respectively, outperforming an average human reader (patient-level sensitivity and specificity of 0.50 and 0.78, lesion-level sensitivity of 0.43 and 0.91 respectively). CONCLUSIONS: AI has the potential to detect and localize csPCa on commonly used b-mode TRUS images, enabling the targeting of suspicious lesions during TRUS biopsy without MRI. Download PPT Source of Funding: National Institutes of Health, National Cancer Institute (U01CA196387, to J.D.B., R37CA260346 to M.R.), the generous philanthropic support of our patients (G.A.S.), Depts of Radiology and Urology at Stanford University © 2024 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 211Issue 5SMay 2024Page: e511 Advertisement Copyright & Permissions© 2024 by American Urological Association Education and Research, Inc.Metrics Author Information Indrani Bhattacharya More articles by this author Sulaiman Vesal More articles by this author Hassan Jahanandish More articles by this author Moonhyung Choi More articles by this author Steve Zhou More articles by this author Zachary Kornberg More articles by this author Elijah Richard Sommer More articles by this author Richard E. Fan More articles by this author James D. Brooks More articles by this author Mirabela Rusu More articles by this author Geoffrey A. Sonn More articles by this author Expand All Advertisement PDF downloadLoading ...