Integrating Location and Tissue Type Information in Entropy -Based Coupled Object Segmentation of Brain Structures

We propose a three – dimensional , nonparametric , entropy – based coupled , multi – shape approach to segment subcortical brain structures from magnetic resonance images (MRI) using tissue type and location information It integrals geometrical relationships among different structures into the algorithm by coupling them it defines an entropy –based energy function which is minimized using quasi – Newton algorithm. To this end , the required probability density functions (pdf) are estimated iteratively using nonparametric Parzen window method Also using FANTASM tissue segmentation, a scaled pdf is defined for the tissue type of each structure In addition, based on the training datasets , a pdf is defined for the location of each structure . Results are given for the segmentation of caudate , thalamus putamen, patlidum, hippocampus, and amygdale illustrating superior performance of the proposed method compared to those of previous methods

کلیدواژه ها:

Image segmentation ، brain structures ، location and tissue type information ، medical image processing ، magnetic resonance imaging (MRI)