Classification and Detection of Brain Tumor Using Segmentation Techniques on MRI Images

In health care monitoring and observations, medical imaging is playing a vital role. Intelligent machine learning techniques for computer-aided diagnosis for automatic identification and detection of infected tissues, defected organs require accurate image segmentation algorithms. In medical image analysis, brain tumor is of high priority for the better diagnosis to improve the quality of the treatment, which increases the existing rate of the patient. MR (magnetic resonance) scans organs, bones and tissues by using immense magnetic forces and radio waves to generate images, which gives clear comprehension that provides a powerful medical diagnostic tool especially for soft tissues like brain. It is a challenging and time-consuming process to identify brain tumors from MRI scans manually in medical routing. Hence computational image segmentation involves systematic and more simplifying anatomical organs, which gives complete information of three-dimensional brain tissue and provides accurate volumetric measurements and other regions of interest (ROI). In this paper, the methodology of implementation of three techniques namely K-Means clustering, thresholding and region growing segmentation algorithms for a spontaneous segmentation of white matter (WM), cerebrospinal fluid (CSF) and gray matter (GM) are presented. The add-on cranial regions and the existence of tumors and also described for the performance evaluation of the three different segmentation algorithms for accurate and significant quantitative search for classification, identification and detection in biomedical images. The experiments are carried out on MRI brain images, the classification accuracy on both training and test data images demonstrates the comparative investigation with the other existing machine learning techniques.