Efficacy of Real-Time Computer-Aided Detection of Colorectal Neoplasia in a Non-Expert Setting: A Randomized Controlled Trial

Aims One-fourth of colorectal neoplasias are missed during screening colonoscopies; these can develop into colorectal cancer (CRC). Several deep learning based real-time computer-aided detection (CADe) systems proved their efficacy in improving the performance of expert endoscopists in neoplasia detection. We performed a multicenter, randomized trial to assess the efficacy of a CADe system in detection of colorectal neoplasias in a non-expert setting to challenge the CADe impact in a real-life scenario.

Methods We analyzed data of consecutive 40- to 80-years-old subjects undergoing screening colonoscopies for CRC, post-polypectomy surveillance, or workup due to positive results from a fecal immunochemical test or signs or symptoms of CRC, at 5 European centers from July through September 2020. Patients were randomly assigned (1:1) to groups who underwent high-definition colonoscopies with the CADe system or without (controls). As CADe, we used a convolutional neural network with convolutional and max pooling layers (GI-Genius, Medtronic) that was integrated in the endoscopy system (i.e. real-time output on the same endoscopy monitor). A minimum withdrawal time of 6 minutes was required. The primary outcome was adenoma detection rate (ADR, the percentage of patients with at least 1 histologically proven adenoma or carcinoma). Secondary outcomes were adenomas detected per colonoscopy, and withdrawal time.

Results The final analysis included 660 patients (age: 62.3±10.0 years old; gender M/F: 330/330). ADR was statistically significantly higher in the CADe-group (176/330, 53.3 %) than in the control group (146/330, 44.2 %; OR: 1.44; 95 % CI:1.06 to 1.96), as well as APC (1.26; 95 % CI:1.14-1.38 vs 1.04; 95 % CI:0.93-1.15; incident rate ratios, IRR:1.21; 95 % CI:1.05-1.40). No statistically significant difference in withdrawal time (CADe: 8.1±1.61 minutes vs control: 7.9±1.53; p = 0.06) was observed.

Conclusions In a multicenter, randomized trial, we found that including CADe in real-time colonoscopy significantly increases ADR and adenomas detected per colonoscopy in a non-expert setting.

Citation: Repici A, Spadaccini M, Antonelli G et al. OP4 EFFICACY OF REAL-TIME COMPUTER-AIDED DETECTION OF COLORECTAL NEOPLASIA IN A NON-EXPERT SETTING: A RANDOMIZED CONTROLLED TRIAL. Endoscopy 2021; 53: S6.

Publikationsverlauf

Artikel online veröffentlicht:
19. März 2021