The occipital (visual) cortex of mammals contains topographic maps of the
visual field, such that a stimulus in the shape of an X will produce neural
activity which, when viewed on a flattened representation of the cortex,
also looks like an X. Topographic maps form because it is less metabolically
costly for highly interconnected neurons to lie next to each other on the
cortical surface. Topographic maps therefore reveal important information
about the computations carried out in different parts of the brain.
It has been hypothesized that many other areas, outside occipital cortex,
also contain topographic maps. However, research has stalled because there
are presently no quantitative tools available for assessing topographic
organization. The goal of this project is to produce robust quantitative
tools for determining the presence of topographic maps in a statistically
rigorous manner. Much of the preliminary work has already been done. Matlab
programs already exist which map existing human fMRI data to a
two-dimensional representation of the cortical surface, and which perform
regression analyses, resulting in quiver plots (see image) that can be used
to visualize quantitative measures of topography.
The specific aims of the project are to further develop and validate these
tools. In particular (i) to improve and extend upon the current methods for
visualizing the data and statistical results (ii) to define regions in which
topography is known to exist, and regions in which it is known not to exist,
so providing a method to establish the hit rate and false alarm rate for the
techniques (iii) to examine the hit and false alarm rates of different
statistical techniques, including 'robust statistics' techniques. (iv) to
explore optimal methods for collecting new fMRI data suitable for processing
using this family of techniques.
The project will be conducted using matlab in the laboratory of Anthony
Jack, in collaboration with Daniela Calvetti.
|
