Design and Modelling of VGG-16 Architecture for Lung Tumor Detection

The most common type of cancer, lung cancer, continues to be the leading cause of cancer-related death in all age categories. To aid early detection, a variety of representational strategies for lung nodules have been put forth. However, these methods run into problems since the datasets are too small and not diverse enough. A method to increase the data set and visualize the representation of pneumonic nodules is needed to solve this problem. Forward and Backward GAN (F&BGAN), a unique data expansion method based on Generative Adversarial Networks (GANs), is presented and demonstrated to produce outstanding clinical images. The Backward GAN(BGAN) algorithm is used to examine images, and the Forward GAN (FGAN) technique is used to create alternative images. The multi-scale VGG-16 network additionally provides a framework for successfully removing discriminative components from stacked layers. The Lung Image Database Consortium and Image Database Resource Initiative data set are used to evaluate the proposed method, which shows impressive sensitivity of 91 percent, specificity of 90 percent, and accuracy of 96 percent in the Area Under the Receiver Operating Characteristic Curve (AUROC).