MP43-02 LESSONS LEARNED IN APPLYING DEEP LEARNING TO FACILITATE PROSTATE MR-US FUSION BIOPSY WORKFLOW

Abstract

You have accessJournal of UrologyProstate Cancer: Detection & Screening IV (MP43)1 Sep 2021MP43-02 LESSONS LEARNED IN APPLYING DEEP LEARNING TO FACILITATE PROSTATE MR-US FUSION BIOPSY WORKFLOW Simon John Christoph Soerensen, Richard E. Fan, Wei Shao, Indrani Bhattacharya, Arun Seetharaman, Michael Borre, Alan Thong, Katherine To’o, Mirabela Rusu, and Geoffrey A. Sonn Simon John Christoph SoerensenSimon John Christoph Soerensen More articles by this author , Richard E. FanRichard E. Fan More articles by this author , Wei ShaoWei Shao More articles by this author , Indrani BhattacharyaIndrani Bhattacharya More articles by this author , Arun SeetharamanArun Seetharaman More articles by this author , Michael BorreMichael Borre More articles by this author , Alan ThongAlan Thong More articles by this author , Katherine To’oKatherine To’o More articles by this author , Mirabela RusuMirabela Rusu More articles by this author , and Geoffrey A. SonnGeoffrey A. Sonn More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000002064.02AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Targeted biopsy improves prostate cancer diagnosis. Accurate prostate gland segmentation on MRI is critical for accurate biopsy. Manual gland segmentation performed by urologists or radiologists is tedious and time-consuming. To address this clinical problem, we sought to develop a deep learning model to rapidly and accurately segment the prostate on MRI and to implement it as part of routine MR-US fusion biopsy in the clinic. METHODS: 905 subjects underwent multiparametric MRI at 29 institutions, followed by MR-US fusion biopsy at one institution. A urologic oncology expert segmented the prostate on axial T2-weighted MRI scans for all cases. We trained a deep learning model, ProGNet, on 805 cases. We retrospectively tested ProGNet on 100 independent internal and 56 external cases. We then prospectively implemented ProGNet as part of the fusion biopsy workflow for 11 patients. We compared ProGNet performance to two deep learning networks (U-Net and HED) and radiology technicians. The Dice similarity coefficient (DSC) was used to measure overlap with expert segmentations. The DSC is widely used to evaluate overlap in segmentation tasks; its value ranges from 0 to 1. A DSC of 1 indicates perfect overlap between segmentations, while 0 indicates no overlap. DSCs were compared using paired t-tests. We are expanding our model's use to provide gland segmentations for all biopsy cases as an initial step in the fusion biopsy workflow. RESULTS: ProGNet (DSC=0.92) outperformed U-Net (DSC=0.85, p<0.0001), HED (DSC=0.80, p<0.0001), and radiology technicians (DSC=0.89, p<0.0001) in the retrospective internal test set. In the prospective cohort, ProGNet (DSC=0.93) outperformed radiology technicians (DSC=0.90, p<0.0001). ProGNet took just 35 seconds per case (vs. 10 minutes for radiology technicians) to yield a clinically utilizable segmentation file. Figure 1 illustrates the step-by-step segmentation workflow at our institution. CONCLUSIONS: This is the first study to employ a deep learning model for prostate gland segmentation for targeted biopsy in routine urologic clinical practice, while reporting results and releasing the code online. Prospective and retrospective evaluations revealed increased speed and accuracy. Source of Funding: This work was supported by Stanford University (Departments of Radiology and Urology) and by the generous philanthropic support of donors to the Urologic Cancer Innovation Laboratory © 2021 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 206Issue Supplement 3September 2021Page: e781-e781 Advertisement Copyright & Permissions© 2021 by American Urological Association Education and Research, Inc.MetricsAuthor Information Simon John Christoph Soerensen More articles by this author Richard E. Fan More articles by this author Wei Shao More articles by this author Indrani Bhattacharya More articles by this author Arun Seetharaman More articles by this author Michael Borre More articles by this author Alan Thong More articles by this author Katherine To’o More articles by this author Mirabela Rusu More articles by this author Geoffrey A. Sonn More articles by this author Expand All Advertisement Loading ...