30120
Tract-Based Cluster Analysis: DTI Group Differences in Autism

Poster Presentation
Friday, May 3, 2019: 11:30 AM-1:30 PM
Room: 710 (Palais des congres de Montreal)
P. A. Luque Laguna1,2, R. A. Dallyn1,2, D. G. Murphy1,2 and F. dell'Acqua1,2, (1)Department of Forensic and Neurodevelopmental Sciences, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, United Kingdom, (2)Sackler Institute for Translational Neurodevelopment, Institute of Psychiatry, Psychology and Neuroscience, King's College London, London, United Kingdom
Background:

Previous studies have reported white-matter differences associated with autism on a wide range of anatomical regions using different methods. However, with conventional voxel-wise analysis it is difficult to associate anatomical specificity to changes occurring on a large and spread group of voxels, while the direct use of few selected tractography reconstructions or brain regions may limit the actual interpretation of global changes within the brain.
To investigate difference in microstructural organisation and diffusion properties between ASD and TD subjects we adopted a novel imaging analysis technique. Tract-based cluster analysis (TBCA1) is a method of voxel-clustering analysis that improves the anatomical specificity and the sensitivity of existing methods without being restricted to specific regions. Specifically, TBCA uses the anatomical coherence between white-matter voxels and the connectivity information provided by tracts to inform the voxel-cluster analysis within a non-parametric statistical framework.In this study, we present results using the TBCA approach on diffusion dataset from a large cohort as part of the UK-AIMS consortium (n=122).

Objectives:

To investigate white-matter differences in autism using diffusion imaging data from the UK-AIMS datasets and a new data-driven method of image analysis (TBCA).

Methods:

MRI acquisition: 61 male subjects with autism and 61 matched controls aged 18 to 45 years. Diffusion-weighted MRI data collected with b-value=1300 s/mm2, 32 diffusion weighted directions and 4 b0s volumes and isotropic voxel size of 2.4mm.
Data pre-processing: motion and eddy-current distortion correction were performed using ExploreDTI2. Whole brain DTI maps of Fractional Anisotropy (FA) were computed for each subject. Each FA map was normalised to the FMRIB58FA 1mm template using flirt/fnirt3.
Statistical analysis: non-parametric cluster-level inference analysis was performed on FA voxels defined within a white-matter mask. For comparison, two methods were tested. One approach was based on the traditional cluster-level inference as implemented in SnPM4 to detect significant clusters formed only by adjacent voxels. Method-2, tract-based cluster analysis with TBCA was applied to detect significant clusters formed by voxels anatomically connected and related to each other according to a connectivity template provide connectivity information for each WM voxel (for more details see Luque-Laguna 2018).

Results:

Figure-1-top: SnPM detected a single cluster covering a region white-matter region around the genu and the anterior body of the corpus callosum.
Figure-1-bottom: TBCA consistently detected multiple clusters of voxels belonging to distinct white-matter tracts such as the left arcuate (green), left uncinate (blue), the genu (red) and the anterior body of the corpus callosum (pink). These tracts are consistent and in line with previous studies showing white-matter changes in adults with ASD.

Conclusions:

The new TBCA results show clusters where the diffusion properties of white-matter differ between ASD and TD subjects. These results also show an increase anatomical specificity simplifying the interpretation of these results. Further research is now required to replicate these results on additional dataset and on other quantitative metrics.

REFERENCES

1 Luque-Laguna et al. 2018 ISMRM (Proceedings)

2 Leemans et al. 2009 ISMRM (Proceedings)

3 Andersson et al. 2007 FMRIB

4 Statistical Non Parametric Mapping Toolbox (SnPM). University of Warwick, 2013.

---

See more of: Neuroimaging
See more of: Neuroimaging