Dear Arif, If you are looking for quality evaluation metrics to quantify the final outcome of the segmentation by comparing the segmented and the ground truth segmentation in medical images, please have a look at our website (http://www.medinfo.cs.ucy.ac.cy) or in research gate. 

i.e C. P. Loizou, A review on ultrasound common carotid artery image and video segmentation techniques, Med. Biol. Eng. Comput., 2014.

Dear Muhammad,

you should better define your problem. Image segmentation (in any field), tries to classify pixels or voxels in two or more classes. So, you should find features that represent your classes in the most discriminative way. It is not the same if you want to segment tumors, organs, bones,etc. In some of them you can use texture information, spatial information and so on. So, you should explain better what you want to do exactly..:)

rough fuzzy setz, local clustering are one the prominent for inhomogeneous images....you can also use modified Chan-Vese algorithm.