Back to all papers

Artificial Intelligence for Colorectal Cancer Diagnosis: A Comprehensive Survey of Modalities, Frameworks, and Clinical Tasks.

November 2, 2026pubmed logopapers

Authors

El-Mosalamy B,Balaha HM,El-Moursi S,Gondim D,Lweesy K,Yousaf J,Ghazal M,El-Baz A

Affiliations (3)

  • Department of Bioengineering, J.B. Speed School of Engineering, University of Louisville, Louisville, KY 40292, USA.
  • Department of Pathology and Laboratory Medicine, School of Medicine, University of Louisville, Louisville, KY 40292, USA.
  • Electrical, Computer, and Biomedical Engineering Department, College of Engineering, Abu Dhabi University, Abu Dhabi 59911, United Arab Emirates.

Abstract

The diagnosis of colorectal cancer encompasses nine complementary modalities: stool-based screening, blood-based biomarkers, cross-sectional imaging, endoscopic evaluation, histopathology, liquid biopsy, molecular and genetic profiling, microbiome analysis, and nanotechnology-based diagnostics. This narrative review synthesizes 160 references spanning seven task-specific artificial intelligence domains, including computer-aided detection, optical diagnosis, invasion depth prediction, computational pathology and radiogenomics, metastasis forecasting, neoadjuvant treatment response prediction, and surgical workflow analysis. We propose a conceptual multimodal pipeline comprising data ingestion, preprocessing and harmonization, feature representation, multimodal fusion, forecasting, explainability, and validation. A structured evidence grading framework classifies each clinical task by evidence level, consistency, and clinical readiness. We further catalog public datasets, challenge benchmarks, and five reporting standards: TRIPOD-AI, CONSORT-AI, SPIRIT-AI, DECIDE-AI, and CLAIM. Computer-aided detection and optical diagnosis demonstrate the highest maturity with Level 1 evidence from randomized controlled trials, while computational pathology, radiogenomics, and metastasis prediction remain at Level 2 with retrospective validation (internal or external). Microbiome screening, nanotechnology diagnostics, and surgical phase recognition are exploratory. The principal barriers to clinical adoption are domain shift, data scarcity, calibration, prospective validation, and workflow integration. We conclude that clinically useful artificial intelligence requires robust external validation, uncertainty quantification, explainable outputs, standardized reporting, and a structured translational pathway from retrospective evidence through regulatory clearance to post-deployment monitoring.

Topics

Journal ArticleReview

Ready to Sharpen Your Edge?

Subscribe to join 11k+ peers who rely on RadAISlice. Get the essential weekly briefing that empowers you to navigate the future of radiology.

We respect your privacy. Unsubscribe at any time.