Deep Learning-Based MRI Brain Tumor Classification Using Convolutional Neural Network Model

Abstract

A brain tumor, the cause of deadly cancer disease among all cancers, is diagnosed by uncontrollable cell growth and abnormal brain cell partitioning. The recent progress in Deep Learning (DL) neural network technology aids the health service department in medical imaging to diagnose several death-causing diseases. The visual learning of image recognition manually may result in fault detection. It can be overcome by the most prevalent task of CNN modeling that commonly uses machine learning algorithms. In our paper, the Convolutional Neural Network (CNN) model is designed with data augmentation and image processing techniques to identify the brain MRI scan images into cancerous or non-cancerous and classify various brain tumor types. The performance comparison of our proposed CNN model with a pre-trained VGG-16 network uses the Transfer Learning (TL) method. With a very small dataset, the experimental result shows that our model is very effective at low computational power with less complexity achieves 100% accuracy than VGG-16 with a 96% accuracy model.