An Automatic Brain Tumor Classification Using Deep Learning Based VGG16 Architecture

Abstract

The abnormal tissues in the brain cause brain tumor. It affects the function of nervous system. The cause of brain tumor is still unknown and the severe condition of brain cancer leads to death. So, to reduce the mortality rate the early diagnosis is required. In this study, an automatic brain tumor classification using deep learning based Visual Geometric Group (VGG) 16 is presented.Initially, the input brain images with 256x256 sizes are given to VGG16 model. The VGG16 architectureis uses convolution layers and maxpooling layers.The Rectified Linear Unit (ReLU) is used as activation function in each convolution and maxpooling layer. Sigmoid layer is used for classification of normal and abnormal brain images.The system uses Magnetic Resonance Imaging (MRI) brain images in REpository of Molecular BRAin Neoplasia DaTa (REMBRANDT) database for performance evaluation. The performance of the system is measured in terms of classification accuracy and cross entropy loss. The classification accuracy of brain image classification system is 92.80% by using deep learning based VGG 16architecture.