Brain Tumor Classification and Segmentation Using Deep Learning

Abstract

The brain is human body's most powerful organ and is responsible for regulating and maintaining all the body’s essential life capabilities. Tumors are the outcome of anomalous and uninhibited cell division. A tumor is an aggregation of tissue that is formed by tremendous cell growth, which continues to grow. A brain tumor is produced in the brain itself or is grown and relocated in another place. No distinguishing cause for the growth of brain tumors has been recognized till date. Although tumors are not very common in the brain (worldwide brain tumors account for just 1.8% of total tumors recorded), the mortality rate of malignant brain tumors is very high due to criticality of the organ. Early detection of brain tumor is a difficult job for doctors. In this paper, authors introduced and implemented a method for classifying brain images with magnetic resonance as normal or abnormal. The abnormal images are further segmented to detect the brain tumor. Classification accuracy achieved is 100%. For segmentation, sensitivity achieved is 85%. Segmentation helps physicians to decide the course of treatment.