You have accessJournal of UrologyBladder Cancer: Detection and Screening II1 Apr 2015PD25-04 VIRTUAL 3D BLADDER RECONSTRUCTION FROM WHITE LIGHT CYSTOSCOPY Kristen L. Lurie, Dimitar V. Zlatev, Roland Angst, Jiyang Gao, Sydney Li, Kathleen E. Mach, Audrey K. Ellerbee, and Joseph C. Liao Kristen L. LurieKristen L. Lurie More articles by this author , Dimitar V. ZlatevDimitar V. Zlatev More articles by this author , Roland AngstRoland Angst More articles by this author , Jiyang GaoJiyang Gao More articles by this author , Sydney LiSydney Li More articles by this author , Kathleen E. MachKathleen E. Mach More articles by this author , Audrey K. EllerbeeAudrey K. Ellerbee More articles by this author , and Joseph C. LiaoJoseph C. Liao More articles by this author View All Author Informationhttps://doi.org/10.1016/j.juro.2015.02.1653AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookTwitterLinked InEmail INTRODUCTION AND OBJECTIVES The high recurrence rate of bladder cancer (BC) requires longitudinal surveillance to detect emerging mucosal lesions. Examination of such lesions with white light cystoscopy (WLC), the standard of care, is largely subjective and limited to clinical notes, diagrams, and still images. A visual history of bladder wall appearance to augment the clinical review may aid clinical management. To facilitate comparison of bladder mucosa over time, we developed a tool to transform in vivo WLC videos into 3D bladder models. METHODS Patients scheduled to undergo transurethral resection of suspected BC were recruited as per an IRB-approved protocol. A clinical-grade rigid cystoscope was used to systematically image the entire bladder. WLC videos were recorded at a resolution of 1280 × 720 pixels, followed by immediate camera calibration to eliminate cystoscope-based image distortions. Video data were fed into an automated structure-from-motion algorithm to generate a 3D reconstruction of the bladder. RESULTS WLC videos having an average length of 226 ± 103 sec (median 216 sec) with a 30 Hz frame rate were obtained from 16 patients. Regions imaged included normal mucosa, inflamed mucosa, and low- and high-grade bladder cancer. Adequate reconstruction could be achieved with only 25% of the frames from a video. Optimal reconstruction was achieved from WLC images depicting well-focused vasculature, when the bladder was maintained at constant volume with minimal debris, and when regions of the bladder wall were imaged multiple times. A representative 3D model of the bladder annotated with the WLC images used for the reconstruction is shown in Figure 1. CONCLUSIONS We demonstrated the first 3D bladder reconstructions from intraoperative WLC videos in patients with suspected BC. A key novelty of this work is the ability to perform the reconstruction using videos from clinical procedures collected with clinical hardware, which suggests the potential for rapid translation of the technology. Envisioned uses of the model include the creation of longitudinal, visual medical records to aid perioperative management and long-term surveillance of patients with BC, and the development of an objective cystoscopy evaluation tool for resident physician education. Future incorporation of machine learning could facilitate early detection of BC recurrence. © 2015 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 193Issue 4SApril 2015Page: e561 Advertisement Copyright & Permissions© 2015 by American Urological Association Education and Research, Inc.MetricsAuthor Information Kristen L. Lurie More articles by this author Dimitar V. Zlatev More articles by this author Roland Angst More articles by this author Jiyang Gao More articles by this author Sydney Li More articles by this author Kathleen E. Mach More articles by this author Audrey K. Ellerbee More articles by this author Joseph C. Liao More articles by this author Expand All Advertisement Advertisement PDF downloadLoading ...