LONI DTI Suite Kristi Clark July 30, 2009 DTI: Driving biological principles A) Post-mortem freezing/dissection technique B) Water diffuses faster parallel to WM tracts A) Lateral view of the internal structures of the left cerebral hemisphere (reprinted from, Ludwig E, Klingler J: Atlas Cerebri Humani. Basel, S. Karger, 1956). B) Reinges et al., Eur J Radiol. 2004 Feb;49(2):91-104. Review. Things to know before starting an analysis What is the question that you want to answer with the data Usually you will want to compare two (or more) populations or look at the effect of a continuous variable such as age MD, FA, ,& are all commonly used metrics that reflect underlying white matter integrity From tractography: can compute length, volume, mean FA, etc. What are the directions These are important to determine the preferred direction of diffusion Not usually in the DICOM data, sometimes wrong, so you need to know these objectively When in doubt, ask your local physicist If they are incorrect, the only thing it should affect is tractography & color maps What are the b values (diffusion weighting) This is the strength with which the diffusion sensitizing pulses were applied: important because this affects all calculations, including MD Usually you can find this on the scanner itself Where to find tools http://www.loni.ucla.edu/twiki/bin/view/Main/KristiClark From the website, download: source code, compiled binaries for linux, generic workflows, an example, & modules.pipe LONI DTI Tools Data needs to be in either DICOM or Analyze format (3D or 4D) 4 atlases available, can substitute any atlas Can now do whole-brain fiber tracking once per subject, and there are utilities to then work with the .ucf For example: OR, AND, NOT functions can be either images or coordinates (with radii) Also there are utilities to compute the histogram, and also do a length selector with a min and max number of points Example 1 Dataset 1: DICOM format, 6 directions with 4 repeats, b=1100 Data was acquired in two acquisitions: Each acquisition consisted of 5 non-diffusion-weighted images (b=0) and 6 diffusion-weighted directions collected in pairs of opposite polarity (e.g. (1, 0, 1) was followed by (-1, 0, -1) for a total of 12 diffusion-weighted images + 5 non-diffusion-weighted images = 17 volumes per input directory Target analysis: compute mean FA for the left SLF Method: use the JHU DTI atlas to define the left SLF in a single subject’s native DTI space Example 1: Data processing overview DICOM & text file that 1 Convert imaging data to describes diffusion gradient analyze & correct the vectors and bvalues gradient vectors 2 Align the single subject’s Correct the images for eddy current T1 data to the single induced distortions and motion subject’s DTI data artifacts (make corresponding 4 adjustments to the gradient vectors) Align the reference 3 anatomical image from the atlas to the single subject’s Compute the diffusion tensor T1-data at each voxel, and its associated eigenvectors and 4 eigenvalues & FA Spatially transform 3 the SLF in the atlas to 4 the single subject’s Use the ROI as a native DTI space mask and compute the mean FA Example 1, part 1: convert data and correct gradient table for slice prescription Step 1: Open correct_slice_prescription.pipe Example 1, part 1: convert data and correct gradient table for slice prescription Step 2: delete scandiff module Example 1, part 1: convert data and correct gradient table for slice prescription Total run time: ~3 minutes Example 1, part 2: correct eddy current induced distortions and motion artifacts Step 2: Open dti_eddy_motion_datawith2repeats.pipe Example 1, part 3: compute the tensor & FA Step 3: Open DTI_analysis.pipe Atlases AAL atlas: GM maps: that based on cytoarch boundaries Postmortem GM regions atlas were manually traced on the WM maps: atlas based on histology of the WM JHU_DTI_based atlas: Major white matter tracts from JHU 27repeat single subj.Based on 10 normal controls. Anatomical underlay: colin27T1_seg (post mortem histology) white matter atlas. Anatomical 151x188x154 (1.0x1.0x1.0) Dimensions: underlay: MNI_ss_cubicmm Anatomical underlay: colin27T1_seg Anatomical underlay: ICBM_152_T1 Dimensions: 182x218x182 from 10 subjects: value divided by These are probabilistic maps(1.0x1.0x1.0) Dimensions: 151x188x154 (1.0x1.0x1.0) Dimensions: 181x217x181 (1.0mm x 1.0mm x 1.0mm) 25 is are probabilistic maps fromA, Panzenboeck MM, Fallon Reference: Tzourio-Mazoyer N, had subjects: value voxel. D, that B, Papathanassiou Thesethe number of subjects who Landeau area in the divided by Reference: Wakana S, Caprihan 10 These files were of subjects as Automated anatomical labelling Crivello F, Etard O, Delcroix N. had that area Originally bilateral 25 is the number downloadedwhopart of SPM5. inof quantitative JH, Perry M, Gollub RL, et al. Reproducibility the voxel. of activations methods applied orientation. tractography in in the proper part of SPM5. Originally regions. Already spm using a macroscopic anatomical These files were downloaded asto cerebral white matter. References: Burgel,U. et MRIthe proper orientation.Neuroimage parcellation of the MNI al., single Neuroimage bilateral regions. Already in (2006). subject brain.29, 1092-1105. Neuroimage 2007; 36: 630-44. Burgel,U. et al., see SPM_anatomy_toolbox_Manual_v15.pdf 15: 273-289. 2002. Many references:(1999) Neuroimage. 10(5), 489-99. Atlases are now on the pipeline Anatomical Spatially aligned ROIs Reference Image Example 1, part 4: ROI analysis Step 4: Open atlas_based_ROI_analysis_for_dti_data.pipe Example 1 summary Result for the mean FA for the left SLF: 0.302377 Total run time: >11 minutes per subject if the grid is available Example 1 tips Use find & replace Smartline Tractography options Atlas-based tractography Acknowledgements UCLA This work was supported by the National Institutes of Health through Roger Woods the NIH Roadmap for Medical Research, Grant U54 RR021813 entitled Arthur Toga Center for Computational Biology (CCB). Information on the National Jeffry Alger Centers for Biomedical Computing can be obtained from John Mazziotta <http://nihroadmap.nih.gov/bioinformatics>. Rico Magsipoc Support for this work was provided by a grant from the Human Brain Ivo Dinov Project (Grant Numbers P20-MHDA52176 and 5P01-EB001955), the Shruthi Chakrapani National Institute of Biomedical Imaging and Bioengineering, National Scott Neu Institute of Mental Health, National Institute for Drug Abuse, National Alen Zamanyan Cancer Institute and the National Institute for Neurologic Disease and JD Trout Stroke. For generous support the authors also wish to thank the Brain Mapping Medical Research Organization, Brain Mapping Support Petros Petrosyan Foundation, Pierson-Lovelace Foundation, The Ahmanson Foundation, Zhizhong Liu Tamkin Foundation, William M. and Linda R. Dietel Philanthropic Fund Amanda Hammond at the Northern Piedmont Community Foundation, Jennifer Jones- Simon Foundation, Capital Group Companies Charitable Foundation, JHU Robson Family and Northstar Fund. The project described was supported by Grant Numbers RR12169, RR13642 and RR00865 from the Kenichi Oishi National Center for Research Resources (NCRR), a component of the Susumu Mori National Institutes of Health (NIH); its contents are solely the responsibility of the authors and do not necessarily represent the official views of NCR or NIH.
Pages to are hidden for
"CCB PPT Template - Download as PowerPoint"Please download to view full document