Study of Techniques used for Medical Image Segmentation Based on SOM

In image processing, segmentation is an important technique which is based on the homogeneous features utilized to partition the image into various regions. In Medical field MR images are widely used, but due to its noise, intensity in homogeneity, Partial Volume Effect (PVE) through voluntary and involuntary movement of the patients and equipments the segmentation process is highly complex. White Matter (WM), Grey matter (GM) and Cerebrospinal Fluid (CSF) are the three main tissue segmentation of MR brain image segmentation. The accurate segmentation of brain tissues facilitates the estimation of tissue volume, tumor detection and estimation of volumes of tumor, which is done by making the image smoother and thus easier to measure. In addition this technique facilitates to estimate the Region of Interest (ROI) in an image. Segmentation is mainly classified as supervised and unsupervised and based on these two there have been various techniques developed for the image segmentation. In medical field, the supervised has less demand as it requires prior knowledge from the external entity. On the other hand, unsupervised segmentation provides more accurate result where it does not need any prior knowledge at any time. The well known Self Organizing Map (SOM) segmentation technique is a type of unsupervised clustering technique utilized to make image quite simple and yields significant accurate segmentation results for the MRI images. This survey paper addresses the various existing methodologies for segmentation of MRI images and presents the issues and advantages related to those approaches.