Meeting Banner
Abstract #3585

A Voxel-Based Analysis of SNR Effect on Diffusion Tensor Imaging

Lian Xue1, Liangsuo Ma2, Khader M. Hasan

1University of Texas Medical School at Houston, Houston, TX, USA; 2Department of Psychiatry and Behavioral Sciences, University of Texas Medical School at Houston, Houston, TX, USA

The study of fractional ansiotropy (FA) mean group differences between healthy and patients has been the primary focus of various diffusion tensor imaging (DTI) investigations, although the accuracy of FA needs to studied due to signal-to-noise ratio (SNR) effects in diffusion-encoded measurements, in particular in regions with low anisotropy such as gray matter. However, there is no systematic study of the influence of SNR in DTI on the estimated FA on a voxel-by-voxel basis for all brain regions. This work is the first report of SNR effects using an unbiased voxel based morphometry (VBM) approach. The comparison of diffusion weighted SNR is realized by selecting different icosahedral schemes from same DTI data set to isolate measurement noise introduced by inter-session factors. The normalization for VBM is completed using the DARTEL technique in SPM5. The group comparison of segmented gray and white matter from different encoding schemes is performed on the same healthy adult controls. The VBM results confirm that FA of GM is effected with lower SNR while FA of white matter is more immune to SNR effects. The multi-faceted encoding scheme approach adopted for comparison gives insight on the choice of encoding scheme in DTI experimental design and the analysis of the minimal scan time needed to provide unbiased measurements to SNR effects.