Segmentation of the human hypothalamus on high-resolution MRI images
OA Version
Citation
Abstract
This thesis addresses the segmentation of the hypothalamus on high-resolution MRI images. The hypothalamus is a very small structure located at the base of the brain that contains functionally distinct nuclei. These nuclei are involved in a number of homeostatic functions, including endocrine and autonomic control, metabolic feedback and regulation, as well as functions relating to learning and memory. The hypothalamic nuclei are affected by a wide range of disorders, but two factors have made it challenging to evaluate the structure of the hypothalamus in both health and disease. First, the small size of the hypothalamus makes it difficult to resolve on structural MRI scans. Second, the borders between individual nuclei of the hypothalamus cannot be observed in MRI scans. The goal of this study was to develop, test, and deploy a protocol to segment different regions of the hypothalamus on high-resolution MRI scans. To achieve this goal, we evaluated different types of structural MRI scans to determine which contrast provided an optimal visualization of the hypothalamus and surrounding anatomical landmarks. Next, we used anatomical landmarks to build a segmentation framework for the hypothalamus in the individual brain. As part of this framework, we mapped segmentation units to individual nuclei. We then deployed this framework on five high-resolution MRI scans and adjusted the segmentation protocol to account for anatomical variability. The results of this study provide the first step in a segmentation protocol for the hypothalamus of the human brain, a procedure that will be used to segment the hypothalamus in healthy and diseased brains.
Description
2024