Classification of Lung Cancer using Transfer Learning based Deep Neural Networks

Lung tumors are the leading global cause of lung-related issues. The best treatment prospects lie in early detection, often involving clinicians to examine the scan reports and detect tumors which is time-consuming and manual procedure. Extensive research has focused on computer-aided diagnostic systems for lung tumor identification using diverse imaging techniques. An effective system requires high accuracy, rapid processing, and automation. Developing such an architecture for early lung tumor detection enables timely clinical decisions for disease management and can save the patient’s life. Deep learning procedures are extensively employed in Health sector in the field of Artificial Intelligence. Aim of this research is to compare different deep learning models that detects the tumor from CT images and categorize it in to four different categories. The methodology comprises of pre-processing, model training and performance evaluation. The models used in this work is modified EfficientNetB3, Resnet50, VGG16, DenseNet121 and AlexNet. The models were trained and evaluated on same dataset. Experimental results shows that good accuracy achieved using modified EfficientNetB3 model.