ABSTRACT
Globally, in recent times one of the most common cancers is lung cancer. Lung cancer is considered an important contributor for death with cancer equally in urban and rural areas. Artificial intelligence algorithms can classify nodules from lung CT images as either benign or malignant. Automated computer-aided diagnosis systems using deep learning algorithms, a subset of artificial intelligence, have established outstanding precision in analysis of images for lung cancer detection. From this study, we can say that computer-aided diagnostics systems not only assist radiologists but pathologists too during screening of patients suffering from lung cancer. But one problem with deep learning is that it is often considered a “black box.” To mitigate this, we need explanatory artificial intelligence. In our research work, we have implemented a simple CNN model and the InceptionV3 CNN Model for detection of lung cancer on the IQ-OTH/NCCD and Kaggle open database. We have obtained accuracy of 93.44% with the simple CNN model and 88.83% with the InceptionV3 CNN Model. Some methods, like screening with low-dose chest CT and pulmonary function tests (PFTs), are required for detection of lung cancer in smokers or ex-smokers. The last section of the chapter concludes with the summary of the study.
