Segmentation of Neonatal BrainMagnetic Resonance ImagesUsingExpectation Maximization

Ramani R, S Balaji

This research paper presents a novel approach for the soft tissue segmentation of NB MRI images utilizing the expectation-maximization approach, initially take the MRI for Skull Stripping or hard tissue removal. Once the Skull is Stripped, it is given for noise removal utilizing anisotropic diffusion filter, and then, the image is segmented into soft non-skull tissues like WM,GM,CSF. The segmentation approach here utilized is on the basis of the EM Gaussian tissue mixture model, where the image stats are utilized to appraisement the coefficient of Dice Similitude Coefficient. The coefficient of dice similitude in the neonatal images is 0,832 for WM.

Volume 12 | 06-Special Issue

Pages: 924-930

DOI: 10.5373/JARDCS/V12SP6/SP20201111