ETL-LWNet: An Ensemble of Transfer Learning Based Light Weight Deep Learning Models for Identification of Lung Cancer Type in CT-Scan Images

Lung cancer is nowadays becoming a common disease that is fatal to the life of any human being. However, modern diagnosis systems have significantly improved to detect such disease at an early stage and helping the medical specialists to effectively prescribe the appropriate treatment. However, there exists different categories of lung cancer which require specific diagnosis and treatment. Recent times have witnessed the development of different machine learning and deep learning methods for identifying any Computed Tomography Scan (CT-Scan) image as either normal or affected by any type of lung cancer. However, owing to hardware constraints there is a need to develop lightweight systems that can achieve maximum possible accuracy. Here a new system called ETL-LWNet which is an ensemble of transfer learning based light weight deep learning models, developed for effectively identifying a CT-Scan image as either normal or the type of lung cancer it displays. The proposed system achieves an accuracy of 89.21% for the Chest CT-Scan dataset which is found to be higher than that achieved by some recent methods.