Susanne G. Mueller1, Karl Young, Michael M. Weiner, Kenneth D. Laxer2
1Center for Imaging of Neurodegenerative Diseases, San Francisco , CA, United States; 2Pacific Epilepsy Program, California Pacifc Medical Center
Synopsis: Voxel-based image analysis approaches, demonstrated different, characteristic abnormality patterns different types of non-lesional partial epilepsy. It is unknown to what degree these patterns exist in individual patients. A two-level multi-modality imaging Bayesian network approach is used with the following aims: 1. to characterize the subtype specific structural abnormalities in non-lesional TLE-MTS, TLE-no and FLE using GAMMA. 2. to the second-level Bayesian network which allows for a robust imaging based subtype classification of single subjects by combining the information obtained at the first level. 3. to identify typical, e.g. FLE gray, and atypical patterns in TLE-MTS, TLE-no and FLE.