Early Detection of Lung Cancer using CT scans Deep Learning techniques

Lung cancer has been one of the most prevalent causes of death cases related to cancer in most parts of the world due to late detection of the tumor. When early detection is successful then the survival of the patient is much better. The present paper offers to utilize the concept of deep learning to develop an automated system that would enable the detection of lung cancer at an early stage by using CT scans. The proposed approach involves the use of Convolutional Neural Network (CNN) 3D, which is utilized to capture volumetric features on the lungs and detect whether the tissues have malignancies or not. Resampling of the voxel, clipping of the Hounsfield Unit (HU), data normalization, and lung morphology are some of the preprocessing steps. The trained model uses augmented 3D patches that were sampled on CT scans, to come up with nodules that are solidly found. Experimental outcomes on the open lung CT dataset have been reported achieving a performance of 96.4 with accuracy, sensitivity, specificity, and AUC of 94.7, 97.2, and 98.1 respectively. These findings indicate that the model has the capability of assisting the radiologist in the due and effective process of diagnosis of lung cancer.