Lung Cancer Classification Model Using Convolution Neural Network

Abstract

Lung cancer has one of the worst mortality rates and incidence rates of any prevalent cancers around the world, and sufferers have a better chance of surviving if the illness is detected early. One of the elements essential to determining the type of cancer is the histopathological diagnosis. Because the type of histology, molecular profile, and stage of the cancer all affect how the disease is treated, it is urgently necessary to analyze the histopathology images of lung cancer. Therefore, deep learning techniques are used to speed up the crucial process of diagnosing lung cancer and lessen the workload on pathologists. It focuses on giving computers the ability to perceive, recognize, and process images in a manner similar to that of human vision and subsequently produce the intended results. It is comparable to imparting human intelligence and instincts to a computer. Deep learning techniques, particularly Convolutional Neural Network (CNN), have improved efficiency in the analysis of cancer histopathology slides. The novelty of this work is to investigate the effectiveness of the proposed model in classifying the digital pathology images for those lung cancer images as either benign or malignant. From the LC25000 dataset, which includes 5000 images for each class, a total of 10,000 digital images were obtained. The histological slides have been divided into benign and malignant cells (cancerous cell squamous) using a shallow neural network structure. The applied model has been attained accuracy ranging from 99.3% to 99.8% in classifying lung malignant from benign lesions. It was experimentally proved that there is no tangible difference in average accuracy between the applied experiments. Therefore, the proposed model has been efficient in comparison with the state-of-the-art.