Automatic Seeded Selection Region Growing Algorithm for Effective MRI Brain Image Segmentation and Classification

摘要

In this paper we proposed automatic seeded point selection region growing algorithm along with clustering technique to solve MRI image segmentation problems more accurately. The manual segmentation, detection and extraction of infected tumor regions of MR image is a tedious job. The accuracy is mainly depends on radiologist knowledge and experience only. The use of computer aided tools is become more choice to overcome the limitations. In this paper, the acquired image is preprocessed by median filter, segmented by automatic seeded region growing segmentation process and the selection of seeded point problem solved. After segmentation, the tumors and their impact analysis can be classified by support vector machine (SVM). Finally from the simulation results the performance accuracies of both benign and malignant tumors compared qualitatively and quantitatively over the existing approaches.