A proposed model for automation of detection and classification of brain tumor by deep learning

Abstract

Brain tumor is the diseases that has true reason of death among the human being. The detection of brain tumor is the initial phase in treatment and is the most exciting work in the medical science. Magnetic Resonance Image of brain is useful in examination of pathology, pre-surgical planning and computer integrated surgery. This paper proposes a model to develop an automatic system which works on Magnetic Resonance images of infected brain to detect and classify brain tumor using deep learning techniques and Convolution Neural Network classifier. This model follows series of processes namely preprocessing, segmentation, feature extraction, tumor detection, tumor classification and tumor location for detecting and classifying brain tumor. The Convolution Neural Network classifier is combined with Gray Scale Co-occurrence Matrix for tumor region extraction, Partial Differential Equation technique for image clustering, K-means and Otsu thresholding method for image segmentation and multi histogram equalization technique for image enhancement. Performance of the proposed system is expected well than other manual and semi-automatic systems that uses Support Vector Machine and Artificial Neural Network classifiers. Accurateness of this system is predicated more than 99%.