Diffusion Tensor Imaging (DTI) for the study of disorders of consciousness
- 1. Diffusion Tensor Imaging (DTI) for the study of disorders of consciousness Stephen Larroque Coma Science Group, GIGA research University of Liège 24/03/2017
- 2. Motivation Connectivity is of paramount importance for consciousness Study connectivity structure (micro and macro) from white matter 2
- 3. DTI preproc in 3 easy steps! (sort of…) 1. Using diffusion magnetic resonance imagery, acquire water molecules (brownian) motion. 2. Estimate tensors ≈ mean motion of water molecules for each brain’s voxel. We get isotropic (round, grey matter) and anisotropic (ellipsoidic, white matter) shapes. 3
- 4. DTI preproc in 3 easy steps! (sort of…) 3. Estimate tractography (=connectivity map): use a probabilistic algorithm (Viterbi) to walk through the tensors and reconstruct a brain connectivity map. 4
- 5. Result 5
- 6. Result 6
- 7. DTI preprocessing theory vs reality DTI preprocessing summary in theory: 1. Acquire DTI images (= hydrogen particles motion) 2. Estimate tensors (= mean particles motion) 3. Tractography (= reconstruct tracts and disambiguate cross-sections) 7
- 8. DTI preprocessing theory vs reality DTI preprocessing summary in practice: 1. Acquire DTI images + T1 2. Reorient both 3. Extract gradients (bvecs and bvals) 4. Brain Extraction (BET) mask on DWI and T1 5. Correct eddy currents 6. Estimate tensors & FA metrics 7. Segment T1 8. Coregister DWI on T1 9. Downsample T1 10. Estimate DWI response function 11. Tractography 12. And more steps depending on your objectives… 8 DTI is still in the process of standardization… but not there yet!
- 9. 2nd-level analysis (group comparison) Fixel-based (local metrics) approach: 1. Normalize all subjects on a (tracts) template 2. Compare locally difference of tracts metrics (eg, AFD for density) Advantage: compare directly the whole structure, but at the expense of losing info at normalization. Connectome approach: 1. Parcellation (Freesurfer) to get regions (or use map provided in MRTRIX) 2. Connectivity matrix (tck2connectome) 3. Graph theory measures and comparison Advantage: respects each subject’s structure and global brain approach, but lose info at parcellation (your analysis is as good as your parcellation) Average/global measures approach: 1. Compute a global measure for each subject (eg, average FA) 2. T-test on the values of one group with the other group 9
- 10. Take home message Enables research of connectivity fibers’ micro- and macro-structure In vivo (and the first one!) By measuring the magnitude and orientation of water diffusion -> non-invasive Useful pre-clinical diagnosis tool Limitation: only ~30% of DTI fibers actually exist in the brain, keep in mind it’s a model! 10
- 11. To go further MRTRIX3 whole documentation Beginner’s DTI preprocessing pipeline (up to connectome analysis): http://community.mrtrix.org/t/beginner-connectome-pipeline-updated/373/2 Fixel-based analysis using MRTRIX3: http://mrtrix.readthedocs.io/en/latest/workflows/fixel_based_analysis.html Connectome analysis using MRTRIX3 (tck2connectome): http://mrtrix.readthedocs.io/en/latest/workflows/structural_connectome.html http://community.mrtrix.org/t/the-output-of-tck2connectome/345 Global measure analysis: see afdconnectivity and http://mrtrix.readthedocs.io/en/latest/workflows/DWI_preprocessing_for_quantitative _analysis.html FSL eddy (eddy currents + motion/realignment correction) Subparcellation Do Tromp’s DTI tutorials, diffusion-imaging.com, 2016 MRTRIX3 community forum! community.mrtrix.com 11
- 12. Thank you for your attention References: •Posterior cingulate cortex-related co-activation patterns: a resting state FMRI study in propofol-induced loss of consciousness, Amico, Enrico, et al, PLoS One 9.6 (2014): e100012. •Multimodal neuroimaging in patients with disorders of consciousness showing “functional hemispherectomy”, Van Someren, E. J. W. (2011), Slow Brain Oscillations of Sleep, Resting State and Vigilance: Proceedings of the 26th International Summer School of Brain Research, Held at the Royal Netherlands Academy of Arts and Sciences, Amsterdam, The Netherlands, 29 June-2 July, 2010, 193, 323. • Neural correlates of consciousness in patients who have emerged from a minimally conscious state: a cross- sectional multimodal imaging study, Carol Di Perri & Mohamed Ali Bahri & Enrico Amico & Aurore Thibaut & Lizette Heine et al., The Lancet Neurology, 2016 •Do Tromp, http://www.diffusion-imaging.com/, 2016 •Amico et al., Conf Proc IEEE Eng Med Biol Soc. 2015 12
- 13. BONUS SLIDES
- 14. How DTI works – A small tale 14 A Ducks Tale about Imagery (DTI)
- 15. How DTI works – A small tale 15
- 16. How DTI works – A small tale 16
- 17. How DTI works – A small tale 17
- 18. How DTI works – A small tale 18
- 19. How DTI works – A small tale 19
- 20. How DTI works – A small tale 20
- 21. How DTI works – A small tale 21
- 22. How DTI works – A small tale 22
- 23. How DTI works – A small tale 23
- 24. How DTI works – A small tale 24
- 25. How DTI works – A small tale 25
- 26. How DTI works – A small tale 26
- 27. How DTI works – A small tale 27
- 28. How DTI works – A small tale 28
- 29. How DTI works – A small tale 29
- 30. How DTI works – A small tale 30
- 31. Duck’s tale imagery to DTI Ducks with GPS = hydrogen particles (in water molecules) Ducks motion = Brownian motion (influenced by environment) Ellipsoid of average travel distance = FA tensors (Fraction Anisotropy) Rivers = white matter fibers tracts Solving river cross-sections = tractography 31