DEVELOPMENT OF A DIGITAL HUMAN BRAIN ATLAS

A digital human brain atlas plays a crucial role in brain mapping for the neuroimaging community. Traditionally, a digital brain atlas contains anatomical information, while structural properties are not presented. Recent development of diffusion tensor imaging (DTI) allows for unique acquisition of information regarding the microstructure of brain tissue. The accuracy of the spatial normalization and subsequent comparisons is contingent upon the use of a DTI template representative of the micro-architecture of the human brain. The previously developed “IIT DTI brain template” was produced in ICBM-152 space. However, low-dimensional registration was used, leading to a mismatch of DTI information across subjects and a mismatch of the anatomy in the IIT and ICBM-152 templates. In this thesis, a significantly improved DTI brain template in ICBM-152 space was developed. The accuracy of spatial normalization in DTI data depends on the quality of data, the effectiveness of the registration technique, and the characteristics of the DTI brain template. Both study-specific and standardized human brain DTI templates exist. The role of both types of templates in spatial normalization warrants further investigation. In this thesis, the effect of standardized and study-specific human brain DTI templates on the accuracy of spatial normalization was investigated. Conventional atlas-based white matter (WM) segmentation is widely used for automated selection of ROI on DTI investigations. However, it suffers from misregistration and inaccurate spatial transformation across subjects. Skeletonized atlasbased segmentation was recently adopted in several DTI studies. However, the use of xiii skeletonized atlas-based segmentation in studies of WM ROIs has not yet been evaluated. In this thesis, the effects of conventional and skeletonized atlas-based segmentation on DTI investigations of WM ROIs were compared. An accurate digital human brain atlas containing different types of high quality MRI data and anatomical labels for both WM and gray matter (GM) in standardized space is desirable for a variety of brain imaging studies. The IIT2 DTI template was developed recently in ICBM-152 space. In this thesis, the quality of the IIT2 template was further enhanced. Furthermore, this publicly available resource was extended into a comprehensive GM atlas of the human brain.PH.D in Biomedical Engineering, May 201