Thank you very much, Tim, for the help. I think I ‘d like to try the wb_command -volume-affine-resample approach. Could you please help me with a complete command given a “mask.nii.gz” that needs to be transformed into a “rfMRI.nii.gz” space.
Thanks, -Heracles From: Timothy Coalson Sent: Monday, March 19, 2018 1:58 PM To: HERACLES PANAGIOTIDES Cc: Glasser, Matthew ; hcp-users Subject: Re: [HCP-Users] Fw: FSL's resampling does not truly respect the coordinates specified by the nifti headers, which makes an FSL solution more complicated. FSLview also doesn't respect those coordinates, and I am unsure whether its "alignment" is the same as its resampling uses, or whether it is incorrect in a different way entirely. FSL's newer viewer may display things correctly aligned, but their resampling still doesn't respect that alignment. I don't know if FSL includes tools to easily deal with these quirks. Workbench does not have these quirks in its resampling, and if your volumes are XYZ ordered and plumb, then what you see is what you get (and there is oblique volume view for any strange volume files). If you are committed to FSL-only solutions, then you may want to try the FSL-specific support instead: https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Support Tim On Sat, Mar 17, 2018 at 8:22 PM, HERACLES PANAGIOTIDES <he...@uw.edu> wrote: Tim, Thank you very much for taking the time to reply. I am hoping to be able to resolve this issue with FSL. I have no experience with the Workbench platform. I can look in that direction if I run out of FSL options. Do you know if there is anyone who might be able to help with this? Thanks again, -hp From: Timothy Coalson Sent: Friday, March 16, 2018 2:15 PM To: Glasser, Matthew Cc: HERACLES PANAGIOTIDES ; hcp-users@humanconnectome.org Subject: Re: [HCP-Users] Fw: Load both of the volumes into wb_view (high-res mask and low-res fMRI - you may want to separate out a single frame from the fMRI to keep memory usage down) and see if they align with each other (if they don't display correctly, turn on oblique volume drawing mode). If they do align, the answer is trivial, use wb_command -volume-affine-resample with the identity matrix as the transform (for smoother edges, use trilinear interpolation of the mask, then threshold it at 0.5 with -volume-math). If not, you will need a transform of some kind, for instance, one that takes the fMRI into MNI space (which would need to be inverted before using it for this). workbench does not have the same issues as FSL with different FOV between different volume files - all alignment is done by millimeters, not by corner of FOV. Tim On Fri, Mar 16, 2018 at 12:58 PM, Glasser, Matthew <glass...@wustl.edu> wrote: Perhaps if this were done with wb_command -volume-resample it would work better, but Tim would have to comment on this. Matt. From: HERACLES PANAGIOTIDES <he...@uw.edu> Date: Friday, March 16, 2018 at 12:11 PM To: Matt Glasser <glass...@wustl.edu> Subject: Re: [HCP-Users] Fw: This is what I used: applywarp -i left_Accumbens.nii.gz -r rfMRI_REST1_LR_hp2000_clean.nii.gz -o left_Accumbens_in_FA.nii.gz The position came our not in the right place. -hp From: Glasser, Matthew Sent: Friday, March 16, 2018 9:39 AM To: HERACLES PANAGIOTIDES Subject: Re: [HCP-Users] Fw: It should work if you use applywarp with no warp or affine. Peace, Matt. From: HERACLES PANAGIOTIDES <he...@uw.edu> Date: Friday, March 16, 2018 at 11:17 AM To: Matt Glasser <glass...@wustl.edu> Subject: Re: [HCP-Users] Fw: Thanks for the reply, Matt. I did not define the problem properly. The original mask is in MNI space (198 X 263 X 212) but does not have the same number of voxels in each dimensions the rfMRI (91 X 109 X 91). Is it possible to transform the the mask to have the same dimensionality (number of voxels in each dimension) as the rfMRI? Best, -hp From: Glasser, Matthew Sent: Friday, March 16, 2018 7:22 AM To: HERACLES PANAGIOTIDES ; HCP Subject: Re: [HCP-Users] Fw: Why can’t you use an MNI space accumbens mask? Peace, Matt. From: <hcp-users-boun...@humanconnectome.org> on behalf of HERACLES PANAGIOTIDES <he...@uw.edu> Date: Friday, March 16, 2018 at 8:04 AM To: HCP <hcp-users@humanconnectome.org> Subject: [HCP-Users] Fw: Could someone please help me warp a mask (left_Accumbens.nii) to an rfMRI in MNI space. I used the following FLIRT command: flirt –in left_Accumbens.nii.gz -ref rfMRI_REST1_LR_hp2000_clean.nii.gz -applyxfm -init MNI2FA.mat –out left_Accumbens_in_FA.nii.gz I am attaching the input, output and mat files. the rfMRI is too big for this email. I am also attaching an image that shows the original ROI (green) and the transformed/output ROI (red). They are not registered correctly. Alternatively, I used the fnirt command with no better luck. fnirt --ref=rfMRI_REST1_LR_hp2000_clean.nii.gz --in=left_Accumbens.nii.gz -- inmask=left_Accumbens.nii.gz --aff=MNI2FA.mat --iout=left_Accumbens_in_FA.nii.gz Any ideas or suggestions? Thanks, -hp _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users
