Abstracts

Rationale: Brain development of children with epilepsy is a topic of increasing interest and concern. In an ongoing study of children with epilepsy serial volumetric MRI scans demonstrate decreased rates of white matter growth [Herman.,P.D et al. Epilepsia 2010 Apr 2]. This investigation address the issue of whether structural connectivity in these children is also abnormal. If so,an important issue is whether deficient connectivity is associated with common neurobehavioral comorbidities in this population including cognitive and behavioral problems? White matter structure is probed noninvasively by diffusion tensor tractography(DTT). DTT is more sensitive to white matter deficiencies than voxel-wise fractional anisotropy (FA) maps because deficiencies enter into DTT as multiplicative factors while they enter into FA maps linearly. We have developed a new form of DTT called W-matrix tractography (WMT) that is more robust than conventional DTT because it allows for multiple branching from every voxel of every fiber tract [Lee.,J.E. et al Neurology 2008; 70(Suppl):A2]. Here we examine the utility of WMT for studying epilepsy in children. Methods: Study participants are 8 to 18 years of age with controls chosen from age-matched first-degree cousins. Participants undergo serial neuropsychological testing and diffusion tensor imaging (DTI). For each subject, a W-matrix is constructed from DTI data. Each element W(i,j) is a product of 3 factors:(1)a product of the FA s of voxels i and j; (2)a structure factor related to the alignment of the first principal DTI eigenvectors from these two voxels;and(3)a distance factor that decreases with distance between the voxels.Each W(i,j) is a number between 0 and 1, with 0 representing no connectivity and 1 reflecting optimal connectivity.WMT can be implemented in two ways.In the first,an arbitrary seed region of interest is chosen. A neural net simulation is then performed with the W(i,j) s interpreted as transition probabilities. The simulation is continued until a stable spatial pattern of activation is obtained. The spatial distribution of this pattern then represents those brain regions that are connected to the seed region of interest. The second approach involves calculating the eigenvalues and eigenvectors of W. The eigenvectors and eigenvalues respectively represent the spatial distribution and strength of brain connectivity. Results: To date,we collected DTI scans from 5 children with epilepsy and 10 controls.With a seed voxel in the splenium midline, W-matrix neural net simulation converges to steady state in about 100 time steps(Fig.1). Spatial distributions at steady state in Fig.2 shows that the seed voxel is connected to fewer other voxels in children with epilepsy compared to healthy controls. Conclusions: W-matrix tractography is computationally feasible in children with epilepsy and controls and promises to provide unique information regarding structural abnormalities. The finding here suggest decreased connectivity in children with epilepsy with seed voxel in the splenium.