Lung Cancer Detection: Classification and Segmentation of CT Images Using 3D CNN

Lung cancer is a condition where malignant unbalanced tumor having uncontrollable growth progresses by invading the nearby healthy cells of lungs and sometimes to other organs. Lung cancer is one of the major causes of cancer-linked deaths. It is also one of the most diagnosed cancers in the world and stands at second position in diagnosed records. Of course, the diagnosis of cancer is confirmed with other clinical tests, which include different sample examinations. However, for people at high risk, CT scan is a better early detection method. But it is observed that for the same CT scan, when marked for potential lung nodules, they vary from radiologist to radiologist. A radiologist goes through a CT scan and gets a 3D view of the lung by envisaging. So, we decided to leverage deep learning models to replace the work that is usually done by humans (radiologists). We trained our model to predict the same, which will help detect and, if present, locate maximum potential nodules. With the 3D CNN technique, we can train the model to extract and highlight the required features in 3D.