Does your fMRI volume line up with your surfaces? Unfortunately different
software packages handle oblique sforms differently and this causes many
incompatibilities. We recommend removing oblique sforms (e.g. by a rigid
6 dof alignment to standard space) prior to running any other processing.
Peace,
Matt.
On 5/16/14, 3:08 PM, "SUBSCRIBE FSL B. Jeong" <[log in to unmask]> wrote:
>Hi Matt,
>
>I thought I would be almost finish line.
>But, it seems that my results from -volume-to-surface-mapping are not
>perfect.
>What I did was,
>1) I create bilateral midthickness surfaces (lh & rh.surf.gii) with
>-surface-average
>2) then, the surface mapping of fMR data was performed to the
>midthickness surface like below;
>wb_command -volume-to-surface-mapping fMRI.nii.gz lh.surf.gii lh.func.gii
>-ribbon-constrained lh.white.surf.gii lh.pial.surf.gii
>3) resample to 32k
>
>On the workbench, values of func.gii were zero on occipito-parietal area.
>I send a figure via e-mail
>
>I'll appreciate for any comment.
>
>Thanks,
>
>Jeong
>
>On Wed, 14 May 2014 09:22:18 -0500, Matt Glasser <[log in to unmask]> wrote:
>
>>1) Load in some matching GIFTI surfaces (i.e. On the 32k mesh). If you
>>don’t have them, create them with wb_command -surface-resample from the
>>native mesh surfaces in a similar way to how you did for the surface
>>timeseries (but use the regular BARYCENTRIC algorithm).
>>
>>2) No, the matrix is 91282 x 195, which is too big for NIFTI-1 along one
>>dimension, so it is wrapped to the dimensions you see.
>>
>>3) For your melodic output, I would set it to 1 and 1, as each map in the
>>dtseries file is a component. Optionally you can convert the file to
>>dscalar and give each map a name (e.g. Component 1, Component 2, etc).
>>
>>Peace,
>>
>>Matt.
>>
>>On 5/13/14, 10:35 PM, "SUBSCRIBE FSL B. Jeong" <[log in to unmask]>
>>wrote:
>>
>>>Dear Matt,
>>>
>>>Great!
>>>Using -cifti-create-dense-timeseries, I got a CIFTI file "dtseries.nii"
>>>which looked similar to HCP subject dtseries file.
>>>Thanks a lot.
>>>I can find signals only at subcortical area via workbench.
>>>I believe the dtseries.nii contains cortical signals.
>>>Question 1: How can I see the cortical signals in surface using
>>>workbench?
>>>
>>>Using -cifti-convert -to-nifti, the CIFTI file was converted to a nifti
>>>file.
>>>The dimension of the nifti file was 32767 x 3 x 1 x 195.
>>>Question 2: does 3 in y dimension mean left, right, and subcortex OR
>>>other?
>>>
>>>Then, melodic command using 2 nifti files produced melodic_IC.nii.gz
>>>(32767 x 3 x 1 x no. of ICs).
>>>To convert back first IC, -from-nifti option was used like below:
>>>wb_command -cifti-convert -from-nifti
>>>/home/projects/bu/marek/New/test/tmp/test1.ica+/IC01.nii.gz
>>>AtlasDenseTimeseries.dtseries.nii IC_01.nii -reset-timepoints 3 0
>>>
>>>Question 3: if TR is 3 sec, is 3 and 0 in -reset-timepoints correct?
>>>
>>>
>>>
>>>My melodic output also showed in subcortical regions when I see it with
>>>workbench.
>>>
>>>I appreciate if I can check my melodic output on surface with workbench.
>>>
>>>
>>>
>>>Thanks,
>>>
>>>Jeong
>>>
>>>
>>>On Mon, 12 May 2014 13:18:51 +0100, SUBSCRIBE FSL B. Jeong
>>><[log in to unmask]> wrote:
>>>
>>>>Hi Matt,
>>>>
>>>>1. Thank you. I had to, but didn't, use a field map. I'll redo my
>>>>preprocessing with a field map.
>>>>2. I guess v3.2.0 is up-to-date(?) one to public (fsl 5.0.6). I'll
>>>>appreciate any information about the latest version.
>>>>3. In terms of my hurdle which -volume-parcel-resampling option
>>>>produced
>>>>4D fMR volume having all 0 valued voxels, could you look at my data?
>>>>(I'll send a email to you including a link for downloading 2mm MNI
>>>>spaced subcortical ROI volume and 4D fMR data (just having 5 volumes
>>>>(11.2Mb) for convenience)
>>>>
>>>>Thank you,
>>>>
>>>>Jeong
>>>>
>>>>On Fri, 9 May 2014 09:31:13 -0500, Matt Glasser <[log in to unmask]>
>>>>wrote:
>>>>
>>>>>Did you correct your data for distortions with a field map?
>>>>>
>>>>>I believe the latest version of FSLView is 4.0.1, but Dave Flitney,
>>>>>the
>>>>>author, could confirm. Old versions may not display NIFTI files
>>>>>produced
>>>>>by Caret or Connectome Workbench correctly because of additional
>>>>>information in the NIFTI header.
>>>>>
>>>>>Peace,
>>>>>
>>>>>Matt.
>>>>>
>>>>>On 5/9/14, 8:24 AM, "SUBSCRIBE FSL B. Jeong" <[log in to unmask]>
>>>>>wrote:
>>>>>
>>>>>>I believe fslview 3.2.0 is mostly recent one.
>>>>>>
>>>>>>bbregister was used to register from fMRI to freesurfer conformed
>>>>>>space.
>>>>>>It seemed be fine, but not perfect especially in the anterior margin
>>>>>>of
>>>>>>caudate.
>>>>>>Which command do you recommend for registration from freesurfer
>>>>>>spaced
>>>>>>fMR data to 2mm MNI space?
>>>>>>I did flirt.
>>>>>>
>>>>>>Thank you,
>>>>>>
>>>>>>Jeong
>>>>>>
>>>>>>
>>>>>>On Thu, 8 May 2014 20:47:25 -0500, Matt Glasser <[log in to unmask]>
>>>>>>wrote:
>>>>>>
>>>>>>>Are you looking at this with an old version of FSLView or something
>>>>>>>like
>>>>>>>that? The images you sent me looked okay (though your EPI is poorly
>>>>>>>aligned with your structural…).
>>>>>>>
>>>>>>>Peace,
>>>>>>>
>>>>>>>Matt.
>>>>>>>
>>>>>>>On 5/8/14, 8:10 PM, "SUBSCRIBE FSL B. Jeong" <[log in to unmask]>
>>>>>>>wrote:
>>>>>>>
>>>>>>>>Not perfectly matched but mostly overlapped between my subcortical
>>>>>>>>ROIs(2mm) and HCP Atlas_ROIs.2.
>>>>>>>>I'll send figures to your MA-TEA.com email.
>>>>>>>>
>>>>>>>>Thanks,
>>>>>>>>
>>>>>>>>Jeong
>>>>>>>>
>>>>>>>>On Thu, 8 May 2014 11:33:35 -0500, Matt Glasser <[log in to unmask]>
>>>>>>>>wrote:
>>>>>>>>
>>>>>>>>>I�m confused again as to where we are because the previous e-mails
>>>>>>>>>are
>>>>>>>>>not
>>>>>>>>>included. Do your ROIs and Atlas_ROIs look okay in FSLView?
>>>>>>>>>
>>>>>>>>>-roi-left and -roi-right you can get from an HCP subject�s
>>>>>>>>>${StudyFolder}/${Subject}/MNINonLinear/fsaverage_LR32k/${Subject}.
>>>>>>>>>${
>>>>>>>>>Hem
>>>>>>>>>is
>>>>>>>>>ph
>>>>>>>>>ere}.atlasroi.32k_fs_LR.shape.gii
>>>>>>>>>
>>>>>>>>>Peace,
>>>>>>>>>
>>>>>>>>>Matt.
>>>>>>>>>
>>>>>>>>>On 5/8/14, 8:07 AM, "SUBSCRIBE FSL B. Jeong"
>>>>>>>>><[log in to unmask]>
>>>>>>>>>wrote:
>>>>>>>>>
>>>>>>>>>>Hi Matt,
>>>>>>>>>>
>>>>>>>>>>Thank you for your comment.
>>>>>>>>>>
>>>>>>>>>>1) Could you any comment in terms of "mostly" correct?
>>>>>>>>>>
>>>>>>>>>>2) Unfortunately, the command below produced a 4D volume which
>>>>>>>>>>have
>>>>>>>>>>all
>>>>>>>>>>zero value:
>>>>>>>>>>
>>>>>>>>>>wb_command -volume-parcel-resampling fMRI.2mm.MNI.nii.gz
>>>>>>>>>>ROIs.2.nii.gz
>>>>>>>>>>Atlas_ROIs.2.nii.gz 2.12 fMRI.subcortex.2.nii.gz
>>>>>>>>>>
>>>>>>>>>>(fMRI.2mm.MNI, ROIs.2, Atlas_ROIs.2 files have same dimension
>>>>>>>>>>(2*2*2mm)
>>>>>>>>>>and same orientation (LAS))
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>3) I can not fully understand the -cifti-create-dense-timeseries
>>>>>>>>>>command.
>>>>>>>>>>what should be placed in -roi-left? How can I make it?
>>>>>>>>>>
>>>>>>>>>>wb_command -cifti-create-dense-timeseries
>>>>>>>>>>AtlasDenseTimeseries.dtseries.nii -volume fMRI.subcortex.2.nii.gz
>>>>>>>>>>Atlas_ROIs.2.nii.gz -left-metric lh.32k.func.gii -roi-left ?
>>>>>>>>>>-right-metric rh.32k.func.gii -roi-right ? -timestep $TR
>>>>>>>>>>
>>>>>>>>>>Thanks,
>>>>>>>>>>
>>>>>>>>>>Jeong
>>>>>>>>
>>>>>>>>——————
>>>>>>>>That seems mostly correct. You¹ll want to download one of our HCP
>>>>>>>>subjects and use
>>>>>>>>${StudyFolder}/${Subject}/MNINonLinear/ROIs/Atlas_ROIs.2.nii.gz as
>>>>>>>>new-parcels and your extracted subcortical labels as cur-parcels.
>>>>>>>>
>>>>>>>>Peace,
>>>>>>>>
>>>>>>>>Matt.
>>>>>>>>
>>>>>>>>On 5/7/14, 5:02 AM, "SUBSCRIBE FSL B. Jeong" <[log in to unmask]>
>>>>>>>>wrote:
>>>>>>>>
>>>>>>>>>Hi Matt,
>>>>>>>>>
>>>>>>>>>1) I started from freesurfer results.
>>>>>>>>>change FS surface and volume to gii or nii
>>>>>>>>>2) create the white and pial surfaces with -surface-average
>>>>>>>>>rh.white.gii lh.white.gii
>>>>>>>>>3) surface mapping of fMR data with -volume-to-surface-mapping
>>>>>>>>>with
>>>>>>>>>-ribbon-constrained option (lh.white.gii lh.pial.gii)
>>>>>>>>>4) make new surface (32k: Sphere.32k.L.surf.gii) with
>>>>>>>>>-surface-create-sphere
>>>>>>>>>5) resampling to a lower resolution (32k) standard mesh with
>>>>>>>>>-metric-resample lh.func.gii lh.sphere.reg.surf.gii
>>>>>>>>>Sphere.32k.L.surf.gii
>>>>>>>>>ADAP_BARY_AREA lh.32k.func.gii
>>>>>>>>>
>>>>>>>>>Then, I did subcortical process.
>>>>>>>>>1) wmparc mgz (1*1*1mm) was resampled to MNI 2*2*2mm space
>>>>>>>>>2) fMR volume also was resampled to this space
>>>>>>>>>3) extracted subcortical labels from wmparc.2.nii.gz with
>>>>>>>>>-volume-label-import
>>>>>>>>>
>>>>>>>>>Now I faced on difficulty in -volume-parcel-resampling
>>>>>>>>>In wb_command -volume-parcel-resampling
>>>>>>>>>I thought cur-parcels may be wmparc.2.nii.gz and new-parcels is
>>>>>>>>>the
>>>>>>>>>extracted subcortical labels from wmparc.2.nii.gz
>>>>>>>>>
>>>>>>>>>But I faced on error message like below:
>>>>>>>>>
>>>>>>>>>parcel volumes are not of type label
>>>>>>>>>
>>>>>>>>>Could you check my process?
>>>>>>>>>
>>>>>>>>>Thanks,
>>>>>>>>>
>>>>>>>>>Jeong
>>>>>>>>>
>>>>>>>>>----------------------------
>>>>>>>>>You need to resample wmparc.nii.gz to whatever resolution you want
>>>>>>>>>the
>>>>>>>>>subcortical data modeled at and then run this command:
>>>>>>>>>
>>>>>>>>>wb_command -volume-label-import
>>>>>>>>>wmparc.${FinalfMRIResolution}.nii.gz
>>>>>>>>>FreeSurferSubcorticalLabelTableLut.txt
>>>>>>>>>ROIs.${FinalfMRIResolution}.nii.gz
>>>>>>>>>-discard-others
>>>>>>>>>
>>>>>>>>>If you want to use the HCP standard 91282 grayordinates space, you
>>>>>>>>>should
>>>>>>>>>download one of our subjects and use
>>>>>>>>>${StudyFolder}/${Subject}/MNINonLinear/ROIs/Atlas_ROIs.2.nii.gz
>>>>>>>>>
>>>>>>>>>You don¹t want to use any -subvolume flag. Use the individual
>>>>>>>>>parcellation (from the command above) as cur-parcels and the file
>>>>>>>>>I
>>>>>>>>>mention from the HCP subject as new-parcels.
>>>>>>>>>
>>>>>>>>>How did you downsample and achieve alignment on the surface? It
>>>>>>>>>might
>>>>>>>>>be
>>>>>>>>>good to recount your steps so I can check them.
>>>>>>>>>
>>>>>>>>>Peace,
>>>>>>>>>
>>>>>>>>>Matt.
|