SURVEY ON VALUMETRIC MEDICAL IMAGE SEGMENTATION TECHNIQUES

Medical image segmentation plays an essential role in the applications such as diagnosis, surgery and radiation treatment planning.  The task is to lable image voxel to a specific anatomic structure so that meaningful assessment like volume and position of the structure can be obtained.  With the requirement of piece-wise continuity, the task is difficult duo to the imaging artifacts.  Classical and recent segmentation approaches will be reviewed in this survey based on the following catalogs:  region based methods that exploit within-region similarities, boundary based methods that exploit between-region differences, hybrid methods that utilize both regional and boundary information and atlas based methods that employ image registration techniques.  The mathematical foundations of these methods as well as their advantages and disadvantages will be discussed.