Identification of Lung Cancer Affected CT-Scan Images Using a Light-Weight Deep Learning Architecture

There has been a growing trend in lung cancer being a major reason for causing fatalities worldwide. Scientists and Medical personals are trying to apply various measures for reducing the adversity of this disease among the patients. It usually requires the understanding of Computed Tomography Scan (CT-Scan) images captured from the lung for inferring if the patient is affected by lung cancer or not. This is essentially a classification task that can be accomplished by typical machine/deep learning models, thereby reducing the human dependency. In this work, a deep learning based method is implemented where, a light-weight Convolutional Neural Network model is trained on a standard dataset IQ-OTH/NCCD which is a collection of CT-Scan images, to identify them as either normal, benign or malignant. Experiments are done on this dataset where the proposed model scored 99% accuracy surpassing that of some recent standard machine/deep learning based methods.