A Novel Weighted Ensemble Model to Classify the Colon Cancer from Histopathological Images

Cancer arises from metabolic irregularities or the convergence of hereditary disorders and poses a life-threatening risk. Colon cancers are the leading cause of mortality in the human population. It is crucial to histologically diagnose malignant cancers to determine the most appropriate treatment for patients. Earlier detection of cancer significantly decreases the risk of death in colon cancers by preventing its advancement within the body. Researchers can utilize machine learning methods and deep learning models to efficiently and quickly examine a larger patient group for cancer detection. This research proposes a new ensemble transfer learning model for the swift detection of colon cancer by integrating and combining different transfer learning models in an ensemble. The proposed ensemble model determines the base learner weight using an innovative strategy rather than manual calculation. The primary purpose of this work is to improve the diagnostic process's overall effectiveness. The findings of this study demonstrate that the proposed approach surpasses current models, making it a valuable tool for clinics to aid medical personnel in more efficiently detecting colon cancer. The average ensemble model performs an accuracy of 98.51%, whereas the weighted-average ensemble model achieves an impressive accuracy of 99.71%, showcasing its superior performance compared to existing approaches.