Brain Tumor Classification using Deep Learning Methods

Abstract

A brain tumor is the most common and destructive disease which takes the patient to the end of life. Thus, treatment planning is a key stage to enhance the quality of patient life. There are various image techniques such as computed tomography, Magnetic Resonance Imaging (MRI), and ultrasound imaging used to measure the Tumor in the brain, lungs, liver, and so on. Classification of Tumor and non-tumor can be done, but it has some limitations, like only a limited number of images can be measured accurately and quantitatively. Therefore, an automatic classification scheme plays an important role in preventing the death rate of human beings, rendering us a challenging task. For this purpose, we take some MRI images to classify Tumor by examining the huge amount of data generated by MRI scan. In this work, brain tumor classification is proposed by extended Convolutional Neural Network (CNN); a deeper convolution layer is designed to improve the performance by using a small kernel size. Surgeons and radiologists can classify brain tumors more easily and effectively using this technique. Extended CNN achieves approximately 99% accuracy rate by showing the experimental results onward.