CAD Scheme Based Brain Lesion Segmentation and Classification Approach

Abstract

Segmentation is a key process in most imaging and classification analysis for ComputerAided Diagnostic or radiological evaluation (CAD). The pixel based method is a key
technique in k-means clustering, as this method is simple and computational complexity is
low compared to other region-based or border-based methods. In addition, segmentation of
biomedical images using the clustering concept as the number of clusters is known from
images of particular regions of human anatomy. The K-means clustering technique is used to
track tumor objects in Magnetic Resonance Imaging (MRI). The key concept of the
segmentation algorithm is to convert an MR input image into a gradient image and then
separate the tumor location in the MR image through the K-media pool. These methods can
obtain segmentation of brain images to detect the size and region of the lesion. Therefore, the
average k cluster can obtain a robust, effective and accurate segmentation of brain lesions in
MRI images automatically and the run time for segmentation of a single lesion is 0.021106.
The detection of the tumor and the removal of the magnetic resonance of the brain are
performed using the MATLAB software. The automatic instrument is designed to quantify
brain tumors using magnetic resonance sets is the main focus of the work. The different
methods used for this concept in the content-based recovery system are precision, memory
and precision value for visual words, descriptive color and border descriptors, diffused
histogram of color and structure. It is expected that the experimental results of the proposed
system will produce better results than other existing systems. Total accuracy of 95.6% is
obtained using GLCM functions in MATLAB software.