Date of Award
Fall 2011
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Biomedical Engineering
First Advisor
Ropella, Kristina M.
Second Advisor
DeYoe, Edgar
Third Advisor
Saad, Ziad
Abstract
Blood oxygenation level dependent functional magnetic resonance imaging has been used extensively for mapping the representation of the visual field within the human brain. Visual field mapping using fMRI has been used clinically to assess patients with cortical pathology and to plan surgical treatment impacting the visual system. The accuracy of fMRI-based visual field mapping methods needs to be better understood for clinical use. This accuracy can be important for presurgical mapping of brain function near a tumor resection site since inaccurate rendition of the underlying neural function could lead to inappropriate resection of viable brain tissue. The most widely used method for visual field mapping is temporal phase mapping. This dissertation investigates the accuracy of temporal phase mapping, specifically focused on the detection of polar angle visual field locations in primary visual cortex. Early studies show that polar angle positions are not uniformly distributed as suggested by animal studies. These non-uniformities are seen as relatively under-represented areas in the visual field maps used to display the fMRI data. This dissertation shows that temporal phase mapping is susceptible to hemodynamic distortions that lead to missassignment [sic] of visual field locations. Further analysis of the non-uniformity in the frequency distribution of voxels representing different angular position within the visual field shows an under-representation of locations near the vertical meridia in V1. These results led to the development of a new retinotopic mapping technique, code-based mapping. The main reason for developing a new retinotopic mapping technique was to reduce the under-representations of vertical meridia posed by using temporal phase mapping when assigning a stimulus location to a voxel. This dissertation shows that code-based mapping is a viable method for mapping visual field locations and produces a uniform distribution of voxels representing different angular positions within the visual field. Furthermore, the code-based mapping method is less susceptible to the hemodynamic biases than temporal phase mapping. With respect to clinical utility of fMRI mapping techniques, the code-based mapping shows a greater potential to accurately map a patient's visual field in the presence of a tumor or other malformations that can induce large noise effects in the fMRI voxel responses.