Segmentation of MR Images using Hybrid kMean-Graph Cut Technique

Abstract

Listen

Translate article

The Image Segmentation is a prominent technique to identify and differentiate the infected and normal regions of the image. The key role of image segmentation comprises extraction of region of interest for better diagnosis of tumors and disease therapy planning. Since, different tumors in brain have diverse shapes, location and intensity value therefore; it is difficult to formulate a general algorithm for image segmentation. Further, the extraction of abnormalities from the brain Magnetic Resonance Imaging (MRI) becomes a challenging task. In this paper, authors have presented the proposed technique for automatic selection of seed points for graph cut segmentation of MR images with tumor. This technique overcomes the common issue of initial seed point selection by exploiting the symmetrical brain structure and using k-mean clustering with graph cut segmentation technique. The results obtained facilitate effective segmentation of infected region.