Classification of Lung Cancer Using Deep Learning

Lung cancer claims a significant amount of lives every year, not only is it tough to treat, the treatment is extremely ineffective and expensive if given in the latter stages of it. This disease claims a great number of valuable lives and is known to inflict one of the most physically painful and mentally bearing deaths. Over the recent years, there has been a lot of research in finding lung cancer in its early stages. In a hope to contribute to this cause, we have developed a method to detect cancerous tissues in a patient’s lungs by using the CT scans images of patients and spot tumors in it. Further, these tissues are classified as benign or malignant. A deep learning model capable of recognizing malignant tissue on CT scan images of a patient. The model has been trained onto a subset of the LIDC-IDRI dataset called LUNA16 that contains 888 images. To these images, image preprocessing techniques have been applied to make it easier for the model to work on. The model utilizes a combination of ResNet and VGG models and gives 98.73% accuracy.