Magnetic Resonance Imaging (MRI) is a powerful technology for imaging the human brain in vivo. However, due to the low resolution of in vivo MRI, around 1 mm, some of the smaller structures, nuclei and tracts, can not be resolved. Even ex vivo MRI imaging, which yields a resolution up to 100 ?m, is not sufficient to accurately segment some of those structures. The brainstem is composed of such small structures and is responsible of the autonomic functions of the body and consciousness and thought to be responsible for a subset of sudden infant death syndrome (SIDS). However, for the infant, the contrast observe in the brainstem is very low do to the lack of myelination. Lack of knowledge about brainstem nuclei and their connections thus limits our ability to understand the mechanism of SIDS and detect the problem early. In this project, we seek to develop atlases and MRI automatic segmentation tool specific to the infant brainstem over the first year of life region using optical coherence tomography (OCT) imaging as training dataset. OCT is a high resolution imaging technique, around 3 ?m, which generate images that contain information comparable to standard histology for neuronal and fiber contents. Contrary to histology, the imaging is performed prior to tissue sectioning, avoiding the distortions induced by histological processing, facilitating the volumetric reconstruction and the accurate registration with MRI. Moreover, our OCT system will be fully automatized. First, we will obtain post mortem brains, both from female and male infants over 4 age categories (3, 6, 9 and 12 months) and image them using MRI scanner, both of the full brain and the excised brainstem. The same brainstem samples will then be imaged by OCT. The optical volumes will be manually labeled for nuclei and tracts and registered to the corresponding ex vivo MRI volumes. Those segmentations will be used to create 4 statistical atlases for the brainstem, for the different age. Finally, the pair OCT-MRI will be used to train algorithm to automatically segment in vivo MRI data using correlation and mutual information metrics. We anticipate that the atlases and the automatic segmentation tool of the brainstem will be of great importance in structural, functional and connectivity studies.
The proposed research is relevant to public health because it will result in an automatic segmentation tool of the infant brainstem, a structure responsible of the autonomic functions of the human body and thought to be responsible for a subset of sudden infant death syndrome. The proposed research is relevant to NIH?s mission to seek fundamental knowledge about the brain and nervous system and to use that knowledge to reduce the burden of neurological disease as well as improve health.
