Behavioral Correlation with Hemispheric Structural Connectivity in Autism

Background: While there have been studies correlating functional connectivity and behavior, structural connectivity analysis for autism has not been widely investigated. Diffusion Tensor Imaging (DTI) provides unique information about the underlying tissue structure of brain white matter in vivo, including both the geometry of fiber bundles as well as quantitative information about tissue properties as characterized by measures such as tensor orientation, anisotropy, and size.

Objectives: The aim of this study is to investigate group differences between correlations of hemispheric structural connectivity (preliminary characterization) in the brain with heart rate variability (HRV) & gaze fixation (GF) for individuals with a diagnosis of high functioning autism spectrum disorder (ASD) and typically developing (TD) individuals.

Methods: The behavioral data was acquired while participants performed an event related facial emotion discrimination task. Happy and fearful faces were presented with a combination of happy and fearful voices. The HRV and GF data was averaged over all the degrees of freedom. Diffusion Tensor Images (with 12 gradient directions) were acquired and spatially normalized using diffeomorphic normalization of tensors into a population specific template. Hemispheric connectivity (h) is characterized as the ratio between the number of fiber tracts crossing mid-sagittal plane and the number of ipsilateral tracts. h=(# intra-hemispheric fiber tracts)/(# inter-hemispheric fiber tracts). The heuristic behind such a characterization is to capture long-range vs. short-range structural connections in the individuals. We used Pearson's correlation coefficients (r) to investigate support for the alternative hypothesis (r_h^ASD != r_h^TD) of finding group differences in the hemispheric ``wiring`` of the brains.

Results: The mean (m) and standard deviations (s) of the hemispheric connectivity for the two groups are: m_h^TD=9.5944, m_h^ASD=9.9749 and s_h^TD=0.9840,s_h^ASD=1.1824. We also performed two-sample t-tests assuming unknown variances (Behrens-Fisher problem) (result, p=0.3363) as well as assuming equal variance (result, p=0.3449). The sample correlations and corresponding p-values for two-tailed distributions for the two measures are presented below:

	Heart Rate Variability (HRV)	Gaze Fixation (GF)
rhTD	0.225187	-0.032184
rhASD	-0.190279	0.030318
z	0.97	-0.13
nTD	14	13
nASD	13	11
p	0.332	0.8966

Conclusions: Although correlation with HRV shows more group difference than that with GF, the difference is statistically insignificant (since p >> 0.05) to support the alternative hypothesis. Although the standard deviation of the hemispheric connecitivity is higher for the ASD group (s_h^ASD>s_h^TD) and null-hypthesis could not be rejected at the 5% significance level, the statistical support for the alternative hypothesis (m_h ^ASD != m_h^TD) is not significant. Preliminary characterization of hemispheric structural connectivity (h) does not reveal any significant structural connectivity differences between ASD and TD individuals. A finer characterization using High Angular Diffusion Imaging (HARDI) might reveal subtle differences between the groups.