Classification of Lung Cancer using Alex-ResNet based on Thoracic CT Images

Abstract

The diagnostic tool can easily identify suspicious shaded areas in CT images from the LIDC-IDRI image repository. The article describes an automated system for detecting nodules within the lungs. An image with a DICOM size of 512 by 512 with filters and segmentation algorithms for identifying the lung area. Moreover, implementations of AlexNet and ResNet-18, 50 and 101 that are fully connected to layers reduce the number of pixels to 227 by 227 and 224 by 224, respectively. Through its performance analyses, extraction of features, classification, sensitivity, specificity, classification, and false alarm rate, the author describes the deep learning network. In deep neural networks, AlexNet had the best results with 100 percent precision and accuracy across SVM multi-class classification compared to ResNet18, ResNet50, and ResNet101.