Photo Credit: romaset

The following is a summary of “Colon Cancer Screening, Surveillance, and Treatment: Novel Artificial Intelligence Driving Strategies in the Management of Colon Lesions,” published in the March 2025 issue of Gastroenterology by Hassan et al.

The integration of Artificial Intelligence (AI) into colonoscopy has significantly transformed the detection and management of colorectal polyps, primarily through Computer-Aided Detection (CADe) and Computer-Aided Diagnosis (CADx) systems. These AI-driven technologies enhance real-time polyp identification and characterization, reducing inter-operator variability and standardizing colonoscopic quality across different levels of expertise. CADe has demonstrated its effectiveness in improving adenoma detection rates, which is critical for reducing the long-term incidence of colorectal cancer. However, despite these advantages, CADe implementation presents certain challenges, including prolonged procedure times, an increase in non-neoplastic polyp resections, and a heightened surveillance burden. On the other hand, CADx shows promise in distinguishing neoplastic from non-neoplastic diminutive polyps, aiding in decision-making regarding polyp resection. Nevertheless, its diagnostic accuracy remains limited, particularly for lesions in the proximal colon, where misclassification rates are higher. Additionally, real-world data reveal a disparity between the efficacy observed in controlled trials and practical outcomes in diverse clinical settings, underscoring the need for further validation studies in broader patient populations. 

Another critical limitation of CADx is its binary classification output and suboptimal specificity, which may reduce clinical confidence in its recommendations. The implementation of explainable AI models could play a pivotal role in increasing physician trust and facilitating the seamless integration of these technologies into routine endoscopic practice. This review comprehensively evaluates the clinical benefits, limitations, and potential risks associated with AI applications in colorectal cancer screening, surveillance, and therapeutic decision-making. Special emphasis is placed on the role of CADe and CADx systems in polyp detection and characterization, along with the existing challenges in translating AI innovations into widespread clinical adoption. Further advancements in AI-driven colonoscopy must address current limitations through improved interpretability, enhanced specificity, and prospective validation in real-world practice to optimize patient outcomes and streamline colorectal cancer prevention strategies.

Source: gastrojournal.org/article/S0016-5085(25)00478-0/abstract