Segmentation of ventricles in Alzheimer mr images using anisotropic diffusion filtering and level set method

Ventricle enlargement is a useful structural biomarker for the diagnosis of Alzheimer’s Disease (AD). This devastating neurodegenerative disorder results in progression of dementia. Although AD results in the passive increment of ventricle volume, there exists a large overlap in the volume measurements of AD and normal subjects. Hence, shape based analysis of ventricle dilation is appropriate to detect the subtle morphological changes among these two groups. In this work, segmentation of ventricle in Alzheimer MR images is employed using level set method and anisotropic based diffusion filtering. Images considered for this study are preprocessed using filters. Anisotropic based diffusion filtering is employed to extract the edge map. This filtering performs region specific smoothing process using the diffusion coefficient as a function of image gradient. Filtered images are subjected to level set method which employs an improved diffusion rate equation for the level set evolution. Geometric features are extracted from the segmented ventricles. Results show that the diffusion filter could extract edge map with sharp region boundaries. The modified level set method is able to extract the morphological changes in ventricles. The observed morphological changes are distinct for normal and AD subjects (p < 0.0001). It is also observed that the sizes of ventricle in the AD subjects are noticeably enlarged when compared to normal subjects. Features obtained from the segmented ventricles are also clearly distinct and demonstrate the differences in the AD subjects. As ventricle volume and its morphometry are significant biomarkers, this study seems to be clinically relevant.