Classification of Brain Tumor Into Four Categories Using Convolution Neural Network

Abstract

A Brain Tumor is a growth of abnormal cells in the brain which can cause discomfort and loss of function for some parts of the brain. The detection and classification of brain tumor using MRI Imaging is done manually by doctors and radiologists which is very time consuming and tedious task as well as the accuracy depends upon the human expertise. So, the use of computer aided technology such as Deep Learning becomes very necessary to overcome these limitations. Our proposed system used Deep Learning algorithm called CNN for automated detection and classification of brain tumor using MRI images into four major categories, Glioma, Meningioma, Pituitary Tumor and No-Tumor. Our dataset consists of 7,183 Brain MRI Images collected from different hospitals and few private scan centers in Mumbai, India. Out of 7,183 Brain MRI images, 5,712 images were used for training and 1,471 images were used for Test Dataset 1 and Test Dataset 2 contains 1,311 images acquired from online sources. Our proposed system has achieved 97.82% accuracy, 97.54% precision, 98.01% recall and 97.76% f1-score on the Test Dataset 1 while Test Dataset 2 achieved 98.70% accuracy, 98.59% precision, 98.66% recall and 98.61% f1-score.