Augmented Bladder Tumor Detection Using Deep Learning

Abstract

Adequate tumor detection is critical in complete transurethral resection of bladder tumor (TURBT) to reduce cancer recurrence, but up to 20% of bladder tumors are missed by standard white light cystoscopy. Deep learning augmented cystoscopy may improve tumor localization, intraoperative navigation, and surgical resection of bladder cancer. We aimed to develop a deep learning algorithm for augmented cystoscopic detection of bladder cancer. Patients undergoing cystoscopy/TURBT were recruited and white light videos were recorded. Video frames containing histologically confirmed papillary urothelial carcinoma were selected and manually annotated. We constructed CystoNet, an image analysis platform based on convolutional neural networks, for automated bladder tumor detection using a development dataset of 95 patients for algorithm training and five patients for testing. Diagnostic performance of CystoNet was validated prospectively in an additional 54 patients. In the validation dataset, per-frame sensitivity and specificity were 90.9% (95% confidence interval [CI], 90.3-91.6%) and 98.6% (95% CI, 98.5-98.8%), respectively. Per-tumor sensitivity was 90.9% (95% CI, 90.3-91.6%). CystoNet detected 39 of 41 papillary and three of three flat bladder cancers. With high sensitivity and specificity, CystoNet may improve the diagnostic yield of cystoscopy and efficacy of TURBT. PATIENT SUMMARY: Conventional cystoscopy has recognized shortcomings in bladder cancer detection, with implications for recurrence. Cystoscopy augmented with artificial intelligence may improve cancer detection and resection.

Keywords: Bladder cancer; Computer-assisted image analysis; Cystoscopy; Deep learning; Diagnostic imaging.

Fig. 1

Representative bladder cancer detection using CystoNet. Green outlines (A–F) represent manual tumor annotation, blue shading (A–C) indicates the algorithm-driven automated tumor segmentation, and red boxes (D–N) indicate the alerts generated by CystoNet. CystoNet tumor segmentation (blue) and manual outline (green) of small papillary tumors located at (A) bladder dome, (B) posterior wall, and (C) anterior wall. The CystoNet alert (red) and corresponding manual annotation (green) shown for (D and E) small solitary tumors and (F) larger, multifocal tumors. In the validation cohort, automated detection of (G) a small tumor at the dome as seen from the bladder neck, (H) a large papillary tumor, and (I) a multifocal papillary tumor with limited background contrast. (J) Example of CystoNet detection of a flat lesion with WLC pathologically confirmed to be carcinoma in situ; (K) another example of carcinoma in situ with (L) corresponding photodynamic diagnosis under blue light cystoscopy. (M and N) False-positive CystoNet alert of a small bladder diverticulum; (O) as the cystoscope is moved closer to inspect the area of the diverticulum, alerting box disappears, which is suggestive of a benign lesion. WLC = white light cystoscopy.