Re: [HCP-Users] Q3 rfMRI

[Glasser, Matthew]

So there is a bug in the .dlabel.nii files that left and right ROIs are not kept separate (because FreeSurfer gives them the same name in both hemispheres).  I have just fixed this for the next data release, but it will have to be worked around here by processing left and right hemispheres separately.  My suggestion for extracting timecourses from the ROIs is to use this approach: wb_command -cifti-create-label ${StudyFolder}/${Subject}/MNINonLinear/fsaverage_LR32k/${Subject}.L.aparc.a2009s.32k_fs_LR.dlabel.nii  -left-label ${StudyFolder}/${Subject}/MNINonLinear/fsaverage_LR32k/${Subject}.L.aparc.a2009s.32k_fs_LR.label.gii -roi-left ${StudyFolder}/${Subject}/MNINonLinear/fsaverage_LR32k/${Subject}.L.atlasroi.32k_fs_LR.shape.gii wb_command -cifti-create-label ${StudyFolder}/${Subject}/MNINonLinear/fsaverage_LR32k/${Subject}.R.aparc.a2009s.32k_fs_LR.dlabel.nii -right-label ${StudyFolder}/${Subject}/MNINonLinear/fsaverage_LR32k/${Subject}.R.aparc.a2009s.32k_fs_LR.label.gii -roi-right ${StudyFolder}/${Subject}/MNINonLinear/fsaverage_LR32k/${Subject}.R.atlasroi.32k_fs_LR.shape.gii For left and right dlabel files, parcellate the dense timeseries to get average timecourses in each ROI: wb_command -cifti-parcellate <dtseries> <dlabel> COLUMN <ptseries> To get the average ROI timeseries out of the parcellated timeseries file: wb_command -nifti-information <ptseries> -print–matrix The alternative, if you were more comfortable with GIFTI files would be to split the CIFTI file into components: wb_command -cifti-separate <dtseries> -metric CORTEX_LEFT <func.gii> -metric CORTEX_RIGHT <func.gii>, would would produce GIFTI timeseries at which point, you could directly use the GIFTI label files.   Peace, Matt. From: basile pinsard <basile.pins...@imed.jussieu.fr> Date: Thursday, December 5, 2013 3:03 AM To: Matt Glasser <glass...@wusm.wustl.edu> Cc: "hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org> Subject: Re: [HCP-Users] Q3 rfMRI Hi Matt, yes I mean homotopic connections. The individual dense connectivity map in the image you sent is not available from the HCP DB, am I wrong? Was this computed from dense timeseries ICA-FIX as provided online using the wb_command -cifti-correlation? Does it add global signal regression to get the negative parts? Maybe the way I extract the signal from cortical ROIS is wrong, but I heavily rely on the cifti header to determine which rows belong to each ROIs so I used the "brain models" to get the IndexOffset and IndexCount of the different brain models, then used the ??????.?.aparc.a2009s.32k_fs_LR.label.gii to get the ROIs in the surface brain models using only the vertices listed in the CIFTI_STRUCTURE_CORTEX_* brain model NodeIndices xml element (which certainly exclude the filled holes in surface and fs medial wall ROI). for instance in the 101915 first scan LR , there are the following brain models: model type            structure                                                                           index    count CIFTI_MODEL_TYPE_SURFACE      CIFTI_STRUCTURE_CORTEX_LEFT        0    29696 CIFTI_MODEL_TYPE_SURFACE      CIFTI_STRUCTURE_CORTEX_RIGHT        29696    29716 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_ACCUMBENS_LEFT        59412    135 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_ACCUMBENS_RIGHT        59547    140 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_AMYGDALA_LEFT        59687    315 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_AMYGDALA_RIGHT        60002    332 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_BRAIN_STEM        60334    3472 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_CAUDATE_LEFT        63806    728 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_CAUDATE_RIGHT        64534    755 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_CEREBELLUM_LEFT        65289    8709 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_CEREBELLUM_RIGHT    73998    9144 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_DIENCEPHALON_VENTRAL_L    83142    706 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_DIENCEPHALON_VENTRAL_R    83848    712 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_HIPPOCAMPUS_LEFT    84560    764 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_HIPPOCAMPUS_RIGHT    85324    795 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_PALLIDUM_LEFT        86119    297 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_PALLIDUM_RIGHT        86416    260 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_PUTAMEN_LEFT        86676    1060 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_PUTAMEN_RIGHT        87736    1010 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_THALAMUS_LEFT        88746    1288 CIFTI_MODEL_TYPE_VOXELS        CIFTI_STRUCTURE_THALAMUS_RIGHT        90034    1248 but the problem might be in the mapping of ROIs from the aparc.a2009s+aseg gifti files. Sorry if it is not clear but I had to reverse engineer the stuff so I may have done it wrong and it might not be possible to do it without some prior hidden in the workbench. Thanks for you insight on this. Basile On Thu, Dec 5, 2013 at 12:52 AM, Glasser, Matthew <glass...@wusm.wustl.edu> wrote: I'm not sure what you mean by homologous connectivity.  Does that mean connectivity between corresponding parts of the hemispheres (like is shown in the attached png from subject 101915—seed vertex is white, corresponding vertex is green)? These data are acquired with a very fast TR (0.720ms), which means that the movement parameters may pick up physiological motion effects in addition to actual subject movements.  In slower TR dat sets these are aliased into the timeseries so you cannot see them.   Peace, Matt. From: basile pinsard <basile.pins...@imed.jussieu.fr> Date: Wednesday, December 4, 2013 4:24 PM To: "hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org> Subject: [HCP-Users] Q3 rfMRI Hi all, trying to process individual rfMRI from HCP, we found unexpected results of connectivity: - almost no homologous connectivity (see attachment) either by taking mean timeseries in aparc.a2009s+aseg rois(as store in fsaverage32k gifti files for surfaces) or by computing the 30gigs voxelwise dense matrix zscore and averaging in the same rois. We did it with and without regressing out mean gray timecourse from the same dense timeseries because it seems to influence largely the connectivity. We also did in on the base preprocessed, and ica-fixed data to check, results are quite similar. Will the pipeline used for preprocessing be released publicly at some point? - the motion estimated is quite shaky and unrealistic (for example the scans of subject 101915 enclosed), do you know which method was used for realignment of the data? If this estimated motion is used both for interpolation of data and regression of nuisance signal, this has certainly huge effect on the connectivity. Was the same data/preprocessing used for estimating the 40subject dense correlation matrix ? This has expected connectivity structure. Thanks for your help. Best. Basile Pinsard. _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users   The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users   The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users