Unravelling the Mysteries of Lung Cancer: Harnessing the Power of Deep Learning for Detection and Classification

Lung cancer is the leading cause of cancer- related death worldwide. Identifying lung nodules is essential in the early detection of lung cancer, which can progress to tumors. Computed tomography (CT) can help with the diagnosis of lung diseases. The primary goal is identifying malignant lung nodules and categorizing lung cancer as benign or malignant. Various image processing, biomarker-based, and machine-automated methods are used to detect lung cancer, but medical practitioners face challenges in achieving accuracy and early diagnosis. This work proposes a novel approach for early and accurate lung cancer diagnosis using the LeNet-5 algorithm. Initially, a median filter is used for preprocessing. Cancerous nodules in lung images were segmented using a hybrid fuzzy with a genetic algorithm (GA). The proposed LeNet-5 is implemented in MATLAB using the dataset obtained by the Lung Image Database Consortium and the Image Database Resource Initiative. Various performance parameters are compared and evaluated with existing algorithms and cutting-edge methods. The experimental outcomes demonstrate that the suggested algorithm has higher classification accuracy (99.2%) than other methods.