Early Detection of Brain Tumors from MRI Using Tensor Flow and Deep Neural Networks

Abstract

Survival rate of brain tumors diagnosed patients is low - around 30 to 36%, that means only two in ten people diagnosed with brain cancer will survive for at least five years, so lakhs of people still dying due to brain tumors other than any other cancer, including thousands of children from India every year. So, early diagnosis plays a pivotal role to save life. Brain tumor classification is difficult because there are 120 types of brain and Central nervous system(CNS) tumors. Accuracy more important in Brain tumors detection, otherwise it can lead to death. One of its applications is the reduction of human judgment in the diagnosis of diseases especially, where minute errors in judgment may lead to tragedy. But, still brain tumor detection is a very big challenge using medical image processing. Recent years, deep learning has dramatically changed and improved the means of recognition, prediction, and diagnosis effectively with high accuracy. This paper focuses on identification of the tumor part from MRI images with the help of Deep Neural Networks(DNN). The size and location of the tumor are as well detected by using CNN and tensor flow. In this proposed work, different angles of brain MRI images taken into consideration then separate networks utilized with various nodes with weight age for the training the segmentation process. For predicting unknown cases, single network formed by utilizing all the trained networks inputs in combined form. The performance evaluation of this paper is to distinguish between normal and abnormal pixels, based on texture based and statistical based features, which is extracted by using GLCM. Promising results with accuracy shows that the performance of the proposed algorithm.