Lung Cancer Detection from LDCT Images Using Deep Convolutional Neural Networks

Lung cancer is second cancer common to men and women as well as it is one of the world's highest cause of death. Reports in recent years have shown that standard X-rays are not effective in diagnosing lung cancer. It has clinically established that low-dose computed tomography (LDCT)-based diagnosis helps to decreases mortality from lung cancer by 20% relative to normal chest X-rays. Deep learning is considered as one of the most beneficial techniques for lung cancer diagnosis. This technique used in many fields, including healthcare, which helps to facilitate complex tasks, analyze medical images, promote reliable diagnosis, and improve diagnostic accuracy. One of the deep learning algorithms is the convolutional neural network (CNN) and in this paper, different deep CNN based models are proposed for lungs cancer detection. The experiments are performed using dataset acquired from Data Science Bowl 2017 (KDSB17). The dataset consists of 6691 LDCT lung images. For testing the efficiency of the model, the accuracy is reckoned, which represents 91.75%. However, due to the sensitivity of this process, other techniques are also used to assess the model's performance including specificity, sensitivity, recall, precision, and f1-score.