In the modern world, the prevalence of cancer has increased due to dietary patterns, lifestyle changes, and, most importantly, the aging of the population. Cancer is characterized by the uncontrolled proliferation of damaged cells. Colon cancer is a type of cancer that starts from the lining of the colon or rectum. Overall, 90%–95% of colon cancers are adenocarcinomas, a type of cancer that originates from the granular epithelium of the lining. Among various cancer types, colon cancer holds a significant rank in occurrence rate since it is the second most diagnosed cancer in women after breast cancer, and the third most diagnosed in men following prostate and lung cancer. The prognosis of colon cancer relies on the extent of metastasis at the time of diagnosis. Early-stage colon cancer often presents subtle and inconspicuous symptoms, leading to delayed diagnosis. Deep learning can be used in various fields such as image classification, speed recognition, and computer vision tasks. Colon cancer is typically diagnosed through a biopsy taken during a colonoscopy. Complementary imaging techniques such as computed tomography (CT) scans are sometimes employed to aid in the diagnostic procedure. Considering the potential benefits, deep learning algorithms have been proposed to assist pathologists in expediting the diagnostic process, thereby potentially reducing mortality rates.

In this chapter, various deep learning and feature extraction methods were investigated to classify images on the histopathological lung and colon cancer dataset (LC2500), including 5000 images of benign colon cancer and 5000 images of adenocarcinomas. The obtained results line with previous research, suggesting the efficacy of deep learning as a promising approach for accurate colon cancer diagnosis.