Interactive lung and colon cancer detection platform

Abstract

Lung cancer and Colon cancer are a leading cause of death worldwide, accounting for nearly 10 million deaths in 2020 which is the second leading death worldwide. Cancer can be reduced through early detection and appropriate treatment. Many cancers have a high chance of cure if diagnosed early and treated appropriately. This paper proposes an easy-to-use interactive Lung And Colon Cancer Detecting Platform (LANCET) to detect lung and colon cancers and communicate the diagnostic results with the users. The proposed system includes a Lung and Colon Cancer Classification Module (LACONIC), a Diagnostic Information and Health Suggestion Providing Chatbot (NINTENDO), and an Easy-To-Use Lung and Colon Cancer Detection Web Application (ANACONDA). For the LACONIC, we resize and split the dataset, and then we use transfer learning to classify lung and colon histopathological images and yield predictions. Using Natural Language Toolkit (NLTK) for the NINTENDO, we train a chatbot that can generate human-like conversation and communicate with users fluently. Finally, the ANACONDA allows patients to select images and get feedback along with suggestions. The systematic tests and validations show that our integrated system is an easy-to-use way of detecting potential cancer and providing health suggestions with high accuracy.