Abstract
The early detection of lung cancer in humans is of significant importance due to its impacts on an individual. The detection of the lung cancer tumor at its early stage is also a challenging problem. Early identification of tumor has the potential to help in saving a large number of human lives. An automatic lung cancer detection and classification (ALCDC) system based on computed tomography (CT) scan images is effective, but the design of a robust lung cancer detection and classification system is a challenging problem. The existing designs of lung cancer detection and classification systems are based on hand-engineered techniques and their outcomes in terms of accuracy and other performance measures are limited. Driven by the exceptional deep learning (DL) success in several recognition related tasks, an ALCDC system based on CT scan images using DL is introduced. For this purpose, using the convolutional neural network (CNN) model, an ALCDC system is built to detect and classify whether the tumors found in the lungs are malignant or benign. The robustness and effectiveness of the proposed ALCDC system is validated using images from the Lung Image Database Consortium (LIDC) and the Image Database Resource Initiative (IDRI). The results indicate that the proposed ALCDC system gives an accuracy of 97.2%. The comparison shows that the proposed ALCDC system performs better than the existing state-of-the-art systems. The proposed ALCDC will be helpful in medical diagnosis research and health care systems.