Human Brain Tumor Classification Based on Deep Learning

Abstract

Classifying brain tumor images is an important part of medical image processing. Helps doctors make accurate diagnoses and treatment plans. Magnetic resonance imaging (MRI) is one of the main imaging tools for studying brain tissue. In this thesis , we propose a method for classifying brain tumor magnetic resonance imaging images using the convolutional neural network of the VGG16 model. Our method integrates a global average pooling layer as an alternative to fully connected layers in a traditional neural network to explore discriminative information. A global average pooling layer can be considered as a solution to the problem of long training time and obtaining the best possible performance from the VGG16 model.The datasets include data compiled from 233 and 73 patients with a total of approximately of 3264 and 253 images from Kaggle for the first and second datasets, respectively. He proposed network structure achieves a significant performance with the best overall accuracy of 98.13% and 98.7%, respectively, for the two studies. The evaluation results demonstrate that our method is effective for brain tumor MR image classification, and it could outperform other comparisons