Leveraging Deep Learning Techniques for Colon Cancer Classification from Histopathology Images

Colorectal cancer is a common and deadly disease due to its complexity and growth. Early detection is crucial for prevention. To detect it early, tissue samples are examined under a microscope, focusing on gland structures. This has resulted in the creation of various evaluation tests to analyze images of polyps and detect colon cancer. In this study, we present an approach to identify cancer in the colon using histopathology images with deep learning algorithms, images include adenocarcinoma and benign tissues. We utilized deep learning models like CNN, ResNet101, EfficientNetB0, InceptionResNetV2, and DenseNet121 to identify colon cancer. These models were evaluated by performance metrics like Accuracy, Error rate, F1-Score, MCC, Kappa, and G-mean. InceptionResNetV2 outperforms very well with an accuracy of 97.25%, The developed model shows potential as a valuable tool to help pathologists make decisions for timely diagnoses. By integrating this model into clinical settings, it can enhance diagnostic workflows, reduce human error, and facilitate timely treatment for patients with colorectal cancer.