Clinical implications of computer-aided real-time size estimation of colorectal polyps during colonoscopy: a prospective study

Accurate polyp size estimation during colonoscopy is crucial for clinical decision making, follow-up, and implementation of cost-saving strategies. Objective sizing methods are lacking, and interobserver variability is high. This prospective, multicenter, study evaluated the accuracy of a novel artificial intelligence (AI)-based algorithm for polyp size estimation.Patient aged ≥18 years undergoing colonoscopy for colorectal cancer (CRC) screening or surveillance were enrolled across three centers. Polyp size was initially assessed by operators using forceps/snare comparison (ground truth). Procedures were recorded, and AI-based polyp size estimates were obtained offline. The primary outcome was AI accuracy in size class determination (diminutive ≤5 mm, small 6-9 mm, large ≥10 mm). Secondary outcomes included size estimation in mm and impact on clinical management strategies.Among 465 polyps (307 diminutive, 107 small, 51 large) from 217 patients (mean age 61.9 [SD 10.4] years, 51.6% female), AI accuracy for size class determination was 85.8% (95%CI 82.5-88.8). Accuracy for diminutive, small, and large polyps was 93.3%, 74.6%, and 55.1%, respectively. The AI tool assigned 90.8% of patients to correct surveillance intervals and achieved mean absolute error of 1.13 mm and root mean square error of 1.40 mm for polyps ≤10 mm.The AI model performed similarly to expert endoscopists in clinically relevant size-related outcomes, potentially improving the accuracy and efficiency of CRC screening.