A DYNAMIC ALGORITHM TO ISOLATE NON CEREBRAL TISSUES OF BRAIN MR IMAGES

Abstract: In the analysis of brain MRI images, the tissues classification plays a vital role in diagnosing the infants to perform the tissue classification, it is quite important to isolate brain from non cerebral tissues of brain and noise from the magnetic resonance structural image. The MR produces different images of type T1, T2, FLAIR images. To isolate non cerebral structures form these different types of image is a challenging task. In this study, we have developed an algorithm for stripping of the non cerebral tissues and noise from the MR images. To remove non cerebral tissues such as skin, skull, fat, muscle and connective tissues which are not region of interest is also most popularly known as skull stripping. The proposed algorithm is stated as Intensity Histogram based quasi anistropic diffusion method (IHQAD). This method is compared to Jaccard similarity coefficients. The method tends to perform better in comparison with existing methods like free surfer, BSE and BET