Validation of Structural/Functional MRI Localization

总结

Summary This proposal on "Validation of Structure and Function in Computational Functional Anatomy" requests four years of continued funding for grant 1R01-EB00975-01. The long-term goal continues to develop Computational Anatomy (CA) methods for assigning functional MRI (fMRI) activity to anatomical coordinates. A central issue in fMRI research is the problem of precisely localizing regions of activation and associating these regions with anatomical labels. fMRI data tend to have both a low signal-to-noise ratio and a low spatial resolution compared with structural MRI data. There is also considerable biologically-based individual variability in the shape of the brain that is a significant confounding variable in associating fMRI activity with a specific brain region. One solution to this problem of individual variability is to transform the functional scan coordinates within the individual's structural scan by constraining anatomically the activation for a given individual to that individual's high resolution cortical structure. In the previous grant, this was achieved via the Large Deformation Diffeomorphic Metric Image Mapping (LDDMM Image) algorithm which increased the statistical power of assigning fMRI signals in a region of interests (ROI) such as the medial temporal lobe and the occipital cortex in memory and visual tasks respectively. The first major focus of our proposal is on the direct assignment of functional signals to cortical coordinate systems via ROI-LDDMM by extending our previous work to memory and vision activity in multiple and connected structures. FreeSurfer has emerged as a powerful tool for parcellating and reconstructing multiple cortical and subcortical structures leading to the second major focus of integrating LDDMM with FreeSurfer parcellation. MRIStudio has also emerged as a powerful tool for analyzing white matter anatomy leading to the third focus of enabling LDDMM to register scalar images derived from diffusion tensor imaging (DTI) data to provide greater statistical power in quantification of white matter anatomy. By integrating these innovative CA tools, we propose to significantly expand upon our initial goals via the following interrelated specific aims. Aim 1 will validate ROI-LDDMM and integration of Free Surfer-LDDMM for studying shape and segmentation of multiple subcortical structures. This will permit mapping of subcortical structures such as the thalamus and basal ganglia and will be applied in Aim 3. Aim 2 will validate the integration of LDDMM in MRIStudio for quantifying white matter fiber tracts connecting subcortical and cortical ROIs via multi-channel LDDMM mapping of DTI data. This will permit reliable assessment of white matter integrity in fiber tracts between predefined ROIs and be applied in Aim 3. Aim 3 will validate the reliability of functionally defined ROIs and structural white matter properties between them using (a) visual retinotopic mapping, (b) cognitive tasks, and (c) white matter anatomy between these functionally defined regions using DTI. The validated tools will be disseminated to the neuroimaging community under the auspices of the Biomedical Informatics Research Network (BIRN) via the CA portal for LDDMM and MRIStudio.