Non-Invasive Segmentation of the Lateral Geniculate Nucleus

The human subcortex contains multiple nuclei that govern the transmission of information to and among cortical areas. In the visual domain, these nuclei are organized into retinotopic maps. Because of their small size, these maps have been difficult to precisely measure using phase-encoded functional magnetic resonance imaging, particularly in the eccentricity dimension. Using instead the population receptive field model to estimate the response properties of individual voxels, I was able to resolve two previously unreported retinotopic maps in the thalamic reticular nucleus and the substantia nigra. I measured both the polar angle and eccentricity components, receptive field size and hemodynamic response function delay in the these nuclei and in the lateral geniculate nucleus, the superior colliculus, and the lateral and intergeniculate pulvinar. The anatomical boundaries of these nuclei were delineated using multiple averaged proton density-weighted images and were used to constrain and confirm the functional activations. Deriving the retinotopic organization of these small, subcortical nuclei is the first step in exploring their response properties and their roles in neural dynamics. I then extended the spatial pRF model to include model parameters for capturing temporal tuning properties of the LGN. The LGN is a laminar structure whose layers can be functionally differentiated based on the response properties of the neurons contained therein. This new spatiotemporal pRF model was designed to detect the differences in the temporal frequency tuning and neural discharge pattern of neurons among the magnocellular and parvocellular layers of the LGN. I then conducted two follow-up experiments where I functionally segment the layers of the LGN using a model-free data-driven approach. I compare the results of the pRF modeling to these model-free data-driven segmentations