Answers are grounded in this paper and cite their sources.
Colorectal polyp sizing is critical for determining optimal resection techniques and surveillance intervals. However, conventional visual estimation and tool-assisted methods remain limited by subjectivity and technical constraints. To overcome these challenges, researchers are actively developing and evaluating artificial intelligence (AI)-based systems for endoscopic polyp sizing. AI-based and computer-assisted sizing methodologies have evolved from distance–magnification correction and an early Argus AI system to anatomical landmark-based measurement using bifurcation-to-bifurcation distance, reference-assisted approaches using forceps or an auxiliary waterjet, and reference-free computer vision approaches, including structure from motion, convolutional neural network (CNN)-based classification, depth estimation models, and SegFormer-based models. More recently, evaluations of CNN-based sizing tools using real-world colonoscopy videos have begun to link automated size estimation with surveillance-related decision-making. Despite these technological strides, current evidence is largely limited to phantom studies, retrospective analyses, single-center studies, early prospective validations, and heterogeneous reference standards. Thus, large-scale multicenter validation studies and real-world evaluations of clinical utility, workflow integration, cross-platform compatibility, regulatory readiness, and cost-effectiveness are required. Ultimately, AI-driven polyp sizing is poised to transform colonoscopy from qualitative observation into a standardized, quantitative clinical decision-support framework.
No takes yet. Share an insight, caveat, or question.
Kim et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: