Deep learning in lower gastrointestinal cancer detection: Advances in endoscopic, radiologic, and histopathologic diagnostics.
Authors
Affiliations (9)
Affiliations (9)
- Department of Internal Medicine, Dayanand Medical College and Hospital, Ludhiana 141001, Punjab, India.
- Department of Internal Medicine, Shrimati Kashibai Navale Medical College and General Hospital, Pune 411041, MahÄrÄshtra, India.
- Department of Medicine, MetroHealth Medical Center, Cleveland, OH 44109, United States.
- Department of Gastroenterology, Dayanand Medical College and Hospital, Tagore Nagar, Ludhiana 141001, Punjab, India.
- Department of Internal Medicine, Trident Medical Center, Charleston, SC 29405, United States.
- Department of Internal Medicine, Louisiana State University Health Shreveport, Shreveport, LA 71103, United States.
- Department of Foundational Medical Studies, Oakland University William Beaumont School of Medicine Rochester, Rochester, MI 48309, United States.
- Department of Anesthesiology, All India Institute of Medical Sciences, New Delhi 110029, Delhi, India.
- Department of Internal Medicine, Cleveland Clinic Akron General Hospital, Akron, OH 44308, United States. [email protected].
Abstract
Gastrointestinal (GI) cancers, particularly colorectal cancer, continue to be a major contributor to global cancer-related morbidity and mortality. Despite significant advancements in screening protocols and treatment strategies, early detection remains a clinical challenge due to the limitations of conventional diagnostic tools, which often suffer from inter-observer variability, limited sensitivity, and time-intensive procedures. In recent years the integration of artificial intelligence (AI), especially deep learning (DL) techniques, into medical diagnostics has opened new frontiers for enhancing detection accuracy, speed, and consistency across clinical domains. This review explores the transformative impact of DL-based AI models in detecting lower GI cancers, focusing on three key diagnostic modalities: Endoscopy; radiology; and histopathology. In endoscopic practice convolutional neural networks are used to detect and classify colorectal polyps in real-time, significantly reducing miss rates and aiding non-specialist endoscopists in decision-making. In radiology DL algorithms trained on computed tomography and magnetic resonance imaging data are valuable for automated lesion detection, segmentation, and staging, often outperforming conventional imaging. Histopathological analysis, traditionally reliant on manual examination, is now accelerated by DL models capable of processing whole-slide images to identify architectural distortions and cellular anomalies with high reproducibility and diagnostic accuracy. This review evaluates DL model performance, including sensitivity, specificity, and area under the curve and addresses technical and ethical challenges, including dataset diversity, interpretability, and integration into healthcare workflows. Ultimately, the convergence of AI and clinical medicine has the potential to improve diagnostic outcomes and personalized care for patients with lower GI cancers.