An Investigation of the Microstructural Organisation of the Fronto-Parietal Branches of the Superior Longitudinal Fasciculus Using Constrained Spherical Deconvolution Based Tractography in Autism Spectrum Disorders

Background:  

Constrained spherical deconvolution (CSD) tractography can facilitate fiber tracking through complex neural regions and generate robust tract reconstructions which cannot be achieved using the diffusion tensor model. Primate and post-mortem research has established that the superior longitudinal fasciculus (SLF) is a white matter tract largely composed of three fronto-parietal longitudinal pathways, the SLF I, II and III. It has been suggested that the SLF I sub-serves the dorsal attention network (DAN), the SLF III sub-serves the ventral attention network (VAN) and the SLF II facilitates communication between these networks. The branches of the SLF have never been isolated in an ASD population. Investigating each distinct branch is important for understanding the neural correlates of attention dysfunction in ASD. As the process of attention is lateralised, evaluating the symmetry of the SLF is also crucial for understanding impaired attention in ASD. 

Objectives:  

The main objective of this research is to use constrained spherical deconvolution (CSD) based tractography to isolate the bilateral SLF I, II and III and investigate diffusion measures of fractional ansiotropy (FA) and the Westin measures of linear and planar diffusion coefficients (CL and CP). The study will also evaluate the symmetry of the SLF I, II and III. 

Methods:  

High angular resolution diffusion imaging (HARDI) data (61 directions, b-value = 1500 s/mm²) was acquired for 45 cases and 45 controls. Preprocessing was completed using ExploreDTI software (http://www.ExploreDTI.com). Data quality checks were performed and subject motion and eddy current induced geometric distortions were corrected for in one interpolation step to minimise blurring effects. The B-matrix rotation was also performed to maintain orientation of the data. The tensor model was applied to the data using robust estimation of tensors by outlier rejection (RESTORE) method. CSD tractography was then performed and the SLF I, II and III were isolated (see Figure 1). FA, CL and CP measures were extracted and independent t-tests were completed. All statistical analyses were Bonferroni corrected at a significance level of p < 0.05/3 = 0.0166

Results:  

In the left SLF I, the ASD group showed greater CL (F (1, 88) = 9.204, p = 0.003) and a strong trend towards greater FA (F (1, 88) = 5.772, p = 0.018) relative to the control group. In the right SLF II, the ASD group also showed greater FA (F (1, 88) = 7.221, p = 0.009) and greater CL (F (1, 88) = 7.862, p = 0.006) than controls. In the SLF II, the ASD group had significantly greater right lateralisation of FA (F (1, 88) = 8.792, p = 0.004) and borderline significance in CL (F (1, 88) = 5.899, p = 0.017) in comparison to the control group.

Conclusions:  

Abnormal structural connectivity of the SLF I, II and III was described thus further substantiating the theory of disrupted cortical connectivity in ASD. The branches of the SLF have been associated with the dorsal and ventral attention networks thus it is reasonable to suggest that aberrant structural connectivity may underpin attentional deficits in ASD.