Abstract: Colorectal cancer represents a globally prevalent malignancy with significant mortality, where current challenges persist in suboptimal sensitivity of early screening modalities and prohibitive costs of precision diagnostics. Artificial intelligence （AI） leveraging deep learning and machine learning architectures have demonstrated transformative potential in endoscopic interventions. In colorectal polyp characterization and adenoma detection, AI-augmented systems significantly enhance adenoma detection rates, reduce miss rates, and optimize quality assurance protocols during colonoscope withdrawal phases. The integration of chromatin structural profiling, radiomic feature extraction, and histopathological analytics enables AI systems to improve diagnostic sensitivity/specificity for premalignant lesions while facilitating lymphatic invasion assessment and TNM staging predictions. Within inflammatory bowel disease management, AI-driven computational models achieve superior quantitation of endoscopic severity indices in ulcerative colitis and demonstrate enhanced discriminative capacity for stricturing phenotypes and penetrating lesions in Crohn's disease compared to conventional clinician evaluation. For advanced therapeutic endoscopy, AI-powered computer vision systems enable real-time procedural phase recognition and automated specimen margin mapping during endoscopic submucosal dissection, thereby improving en bloc resection rates and operative safety profiles. Notwithstanding these advancements, persistent limitations including training dataset heterogeneity and insufficient generalizability. Future development trajectories involving high-quality data and algorithmic optimization are anticipated to drive paradigm shifts toward AI-enabled precision coloproctology.

Key words: Artificial intelligence, Colorectal cancer, Endoscopy, Polyps, Inflammatory bowel disease