Norbert W. Schuff1, 2, Yu Zhang1, Karl Young, Howard Rosen, Michael W. Weiner1
1UCSF, San Francisco, CA, United States; 2VAMC, San Francisco, CA, United States
The best approach for maximizing information from diffusion tensor imaging (DTI) is a matter of intense debate, especially when it comes to quantifying the integrity of nerve fiber bundles. We explored the value of information theoretic measures, such as Kullback-Leibler divergence, for quantifying DTI variations along fibers as indeces of fiber uniformity. Simulations suggest that the approach is useful for capturing global and local features. We also applied the approach to experimental DTI data from patients with cognitive impairments, including Alzheimers disease (AD). The results are consistent with the idea that AD is associated with diminishing structures within fiber bundles.