Genetic Analysis of High Angular Resolution Diffusion Images (HARDI)

Imaging genetics is a new field that extends methods from quantitative genetics to handle brain images. Its goal is to identify features of the brain that are genetically influenced, and then find specific variations at the genomic level that contribute to them. Recent studies using diffusion tensor imaging (DTI) in twins have discovered specific genomic variations that influence standard DTI-derived measures, such as fractional anisotropy (FA). As these associations require hundreds of scans to detect, genetic analyses will be accelerated if some measure could be derived from diffusion images that is more highly genetically influenced than FA. To test this, we scanned 116 young adult twins (29 identical and 29 fraternal twin pairs) with high-angular resolution diffusion imaging (HARDI) at 4 Tesla. We fitted tensor distribution functions (TDFs) to the full HARDI signal at each point in the brain. We then computed TDF-FA, a fiber-crossing/mixing corrected FA using the TDF, and the tensor orientation distribution (TOD), whose spherical integral encodes the volume fraction of fibers detected in each direction. We fitted structural equation models voxel-wise, using multivariate correlation for genetic analysis of vector-valued measures. Fiber directions, expressed using the 3-vector of TOD projections along each axis, were more highly heritable than standard scalar anisotropy measures (e.g., standard FA). FATDF gave higher inter-twin correlations than standard FA, avoiding errors where fibers cross or mix. Due to their higher heritability, these HARDI measures show promise for genome-wide association studies of diffusion images, and are expected to accelerate the search for specific genetic causes of variations in fiber integrity and connectivity in large human populations.