AI-ASSISTED DETECTION, CHARACTERIZATION AND SIZING OF COLORECTAL POLYPS. CAN AI SUPPORT NON-EXPERT ENDOSCOPISTS TO ACHIEVE PIVI THRESHOLDS? INTERIM RESULTS FROM A PROSPECTIVE MULTI-CENTER INTERNATIONAL TRIAL

Aims Real-time in-vivo characterization of colorectal polyps remains limited outside expert centers. Data on AI polyp detection and characterization is promising but accurate sizing remains the missing jigsaw piece. We aimed to study the impact of a novel AI system on non-expert endoscopists' detection, characterization and sizing of colorectal polyps compared to experts.

Methods Prospectively collected endoscopy videos from twelve centers in Europe and Japan were uploaded on a bespoke online platform (Taka-tool). All polyps were histologically proven and sized by three experts. The AI model detects polyps and classifies them as neoplastic/non-neoplastic and diminutive/non-diminutive. We asked Six experts to detect, characterize and size polyps without AI support, and Six non-experts to detect polyps assisted by AI, and to characterize and size polyps without and then with AI.

Results

Table 1

Metric	Non experts+AI	Experts	P value
Sensitivity of characterization on EI	95.5%	92.4%	>0.5
NPV of characterization on EI	90.8%	86.7%	>0.5
Sensitivity of sizing	93.6%	92.2%	>0.5
NPV of sizing	93.1%	92.3%	>0.5

199 videos (100-polyps) were included. On polyp detection, average sensitivity and specificity of non-experts+AI compared to experts was 96.0% and 84.6% compared to 95.7% and 89.9% respectively (p>0.5).Non-experts+AI showed superior sensitivity (95.5% vs 83.3%) and NPV (90.8% vs 70.4%) of characterization on enhanced imaging compared to non-experts alone (p<0.5). On sizing, non-experts+AI achieved accuracy and sensitivity of 84.0% and 93.6%, respectively. Experts’ characterization and sizing metrics were not significantly different from non-experts+AI.

Conclusions This interim analysis suggests our AI system may support non-experts to perform at experts’ level and achieve PIVI-2 threshold (diagnose and leave).Further analysis is underway to understand the impact of the AI system on surveillance interval (PIVI-1).To our knowledge, this is the first report incorporating AI-assisted sizing with detection and characterization.

Publication History

Article published online:
14 April 2022