The role of artificial intelligence in colorectal cancer and polyp detection: A systematic review.

• DOI: 10.1200/jco.2025.43.4_suppl.47
• Publication Date: 2025-01-27T14:31:24Z
• Authors: Sakshi Bai; Bipneet Singh; Jahnavi Ethakota; Oboseh John Ogedegbe; Olanipekun Lanny ...; Akshit Chitkara; Sindhu Rani; Danesh Kumar; JoAnn Hirth; Devin Birsingh Malik

47 Background: Colorectal cancer (CRC) remains a leading cause of cancer-related mortality worldwide. Early detection and accurate diagnosis are essential for improving survival rates optimizing therapeutic strategies. Recently, artificial intelligence (AI) technologies, such as machine learning algorithms, convolutional neural networks (CNNs), computer-assisted diagnostic (CAD) systems, have significantly enhanced traditional tools like colonoscopy histopathology. This systematic review evaluates the current role AI in CRC management, focusing on improvements, predictive modeling, outcome prediction. Methods: A comprehensive literature was conducted using PubMed, Google Scholar, MEDLINE MeSH term to identify studies utilizing with focus imaging, histopathological analysis, modeling. Studies were evaluated accuracy AI-driven systems performance models clinical outcomes. Results: Out 147 identified, 37 met eligibility criteria this review. CNNs, CAD EndoBRAIN showed high polyp classification. Kudo et al. (2020) reported 98% EndoBRAIN. Blanes-Vidal (2019) achieved 96.4% via DCNN. Additionally, Wang (2019, 2020) demonstrated improved Adenoma Detection Rates (ADR) 34.1% 29.1%, respectively, over methods. For prediction, models, including predicted patient prognosis reasonable accuracy, highlighted by Skrede 76% Conclusions: The analysis AI-assisted modalities colorectal reveal promising results, most demonstrating sensitivity, specificity, accuracy. CNN-based particular, show strong potential outcomes screening. included demonstrate that can enhance precision, particularly identifying polyps predicting outcomes, validation rates. However, more external long-term needed fully establish robustness these routine practice.