Advancing Cancer Diagnosis with LungColonGuard: A Deep Learning Framework For Lung and Colon Cancer Detection From Histopathological Image

Genetic mutations or other changes in DNA can contribute to the uncontrolled cell growth that occurs in the human body, known as cancer. Undoubtedly, lung and colon cancers are fatal diseases that contribute significantly to the death and disability of people worldwide. In the initial stage, cancer cells are mostly smaller in size and have not spread to surrounding cells or other organs of the body. Early detection and diagnosis of these forms of cancer can drastically enhance the possibility of successful treatment and a better outcome. The adoption of ML and DL models in computer-aided systems can massively improve the speed and accuracy of cancer diagnosis. It is also beneficial to examine a significantly large number of patients within significantly shorter time periods. In this study, we present a Convolutional Neural Network based deep neural architecture named LungColonGuard which is developed to accurately identify lung and colon cancer. An assessment of comparison is presented between the LungColonGuard model and several transfer learning models, namely: ResNet50V2, ResNet101V2, and ResNet152V2. The proposed model gained the highest accuracy of 99.50% for colon and lung cancer detection.