Analysis of brainstem in Alzheimer MR images using Lattice Boltzmann level set

Abstract

Alzheimer's Disease (AD) is a progressive neurological disorder resulting in the cognitive impairment of elderly people. The brainstem is considered as a significant pathological core for the study of psychological and behavioural symptoms occurring before cognitive impairment. In this work, the brainstem is extracted from T1 weighted brain MR images using Lattice Boltzmann (LB) criterion based regional level set method. The segmentation results are validated using regional statistic and overlap measures. The results show that the region-based segmentation is capable of extracting the brainstem more accurately by preserving the weak and blurred edges of the brainstem. The extracted geometric and textural features are able to show a significant difference in the normal and AD brainstem images. The feature values obtained from AD images are found to be less compared to the normal images. This might be due to the neuronal loss in the brainstem describing the atrophy. Hence, this analysis proved to be clinically significant.