I think this might be way more complicated than is needed. I believe there are structural extended packages that would allow you to run a FreeSurfer command without needing to replicate FreeSurfer files. As for how to get a parcellation from an atlas surface to a subject’s volume, I could tell you how to do that using Connectome Workbench tools, file formats, and our atlas, but I don’t know the details for FreeSurfer.
Peace, Matt. From: Simon Baker <[email protected]<mailto:[email protected]>> Date: Tuesday, November 10, 2015 at 11:34 PM To: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Cc: "Harms, Michael" <[email protected]<mailto:[email protected]>>, Timothy Coalson <[email protected]<mailto:[email protected]>>, Matt Glasser <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] Custom parcellation for HCP data Thanks for the pointers Michael, Tim, and Matt. We've tried a range of approaches but we're still having trouble creating the custom parcellation. Below describes the updated version of our pipeline. Although this version improves the spatial alignment between the custom parcellation and the T1w volume, the quality of the parcellation is still lacking compared to e.g. ribbon.nii.gz. -- 1. Create a high-resolution annotation (parcellation) by randomly parcellating the fsaverage surface into N regions of approximately equal volume. 2. Map the annotation from the source subject (fsaverage) to the target subject using mri_surf2surf. mri_surf2surf --srcsubject fsaverage --hemi xh --sval-annot highres.annot --trgsubject ${SUBJECTID} --srcsurfreg sphere.reg --trgsurfreg /MNINonLinear/Native/${SUBJECTID}.X.sphere.native.surf.gii --tval xh.highres.annot 3. Obtain vertices and faces data from /MNINonLinear/Native/${SUBJECTID}.X.white.native.surf.gii 4. Using the vertices and faces data obtained in step 3 as inputs for the write_surf Matlab function, create the xh.white surface file. 5. Obtain thickness data from /MNINonLinear/Native/${SUBJECTID}.X.thickness.native.shape.gii 6. Using T1wImage = /T1w/T1w_acpc_dc_restore.nii.gz and T1wImageBrain = /T1w/T1w_acpc_dc_restore_brain.nii.gz, we replicate some of the steps found in HCP's FreeSurferPipeline.sh in order to generate /mri/orig.mgz. Specifically: #Make Spline Interpolated Downsample to 1mm Mean=`fslstats T1wImageBrain -M` flirt -interp spline -in T1wImage -ref T1wImage -applyisoxfm 1 -out T1wImage_1mm applywarp --rel --interp=spline -i T1wImage -r T1wImage_1mm --premat=/usr/local/fsl/etc/flirtsch/ident.mat -o T1wImage_1mm applywarp --rel --interp=nn -i T1wImageBrain -r T1wImage_1mm --premat=/usr/local/fsl/etc/flirtsch/ident.mat -o T1wImageBrain_1mm fslmaths T1wImage_1mm -div $Mean -mul 150 -abs T1wImage_1mm mri_convert T1wImage_1mm /mri/orig/001.mgz #Initial Recon-all Steps recon-all -subjid ${SUBJECTID} -sd ${SUBJECTS_DIR} -motioncor -talairach -nuintensitycor -normalization 7. Convert the custom parcellation annotation into a custom parcellation volume using mri_label2vol, with /mri/orig.mgz serving as the output template volume. mri_label2vol --annot xh.custom.annot --temp /mri/orig.mgz --identity --proj frac 0 1 .1 --subject ${SUBJECTID} --hemi xh --o xh_custom_vol.mgz 8. Resample and reslice the custom parcellation volume. mri_convert --resample_type nearest --reslice_like T1wImage xh_custom_vol.mgz xh_custom_vol.nii 9. Combine the resampled and resliced custom parcellation volume from each hemisphere to create a single custom parcellation volume (custom_vol.nii). 10. Find c_ras offset and generate matrix MatrixX=$(mri_info --cras T1wImage | cut -d " " -f 1) MatrixY=$(mri_info --cras T1wImage | cut -d " " -f 2) MatrixZ=$(mri_info --cras T1wImage | cut -d " " -f 3) echo "1 0 0 ""$MatrixX" > c_ras_xfm.mat echo "0 1 0 ""$MatrixY" >> c_ras_xfm.mat echo "0 0 1 ""$MatrixZ" >> c_ras_xfm.mat echo "0 0 0 1" >> c_ras_xfm.mat 11. Invert the c_ras offset matrix convert_xfm -omat c_ras_xfm_inv.mat -inverse c_ras_xfm.mat 12. Apply the inverted c_ras offset matrix to the custom parcellation volume (custom_vol.nii) flirt -in custom_vol.nii -applyxfm -init c_ras_xfm_inv.mat -out parcellation.nii -paddingsize 0.0 -interp nearestneighbour -ref T1wImage 13. [Not sure why, but then we had to] Warp the output from the previous step in order to align the parcellation with T1wImage (i.e., /T1w/T1w_acpc_dc_restore.nii.gz). applywarp --rel --interp=nn -i parcellation.nii -r T1wImage -w /MNINonLinear/xfms/standard2acpc_dc.nii.gz -o parcellation_native.nii 14. See attached screenshot2.jpeg showing /T1w/T1w_acpc_dc_restore, /T1w/ribbon (blue), and parcellation_native (yellow). -- The alignment of parcellation_native is a bit off and it appears to be grainy/pixelated. Do you have any suggestions as to why this is the case? Kind regards, Simon Baker Brain & Mental Health Laboratory Institute of Cognitive & Clinical Neuroscience Monash University On 20 October 2015 at 09:20, Glasser, Matthew <[email protected]<mailto:[email protected]>> wrote: FreeSurfer might do it if you simply specify the correct extension, but if not, mris_convert will do the conversion. Peace, Matt. From: <[email protected]<mailto:[email protected]>> on behalf of Timothy Coalson <[email protected]<mailto:[email protected]>> Date: Monday, October 19, 2015 at 4:44 PM To: "Harms, Michael" <[email protected]<mailto:[email protected]>> Cc: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>>, Simon Baker <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] Custom parcellation for HCP data Additionally, if you generate a cortex-only parcellation, you shouldn't need to put it into a volume to use it with the HCP CIFTI data. However, I'm not sure how to get freesurfer to write GIFTI label files, and unless you generate it on the fs_LR 32k mesh, you'll need to resample it (via registered sphere surfaces). If you can get the label data into matlab, though, you can write it as .func.gii files, and use wb_command to turn it into label.gii, and then to .dlabel.nii for use with CIFTI data. Some care may be needed to keep the right and left labels separate. Tim On Mon, Oct 19, 2015 at 10:08 AM, Harms, Michael <[email protected]<mailto:[email protected]>> wrote: Hi, Did you see this recent thread on the list: http://www.mail-archive.com/hcp-users%40humanconnectome.org/msg01910.html cheers, -MH -- Michael Harms, Ph.D. ----------------------------------------------------------- Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave. Tel: 314-747-6173<tel:314-747-6173> St. Louis, MO 63110 Email: [email protected]<mailto:[email protected]> From: Simon Baker <[email protected]<mailto:[email protected]>> Date: Sunday, October 18, 2015 8:16 PM To: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: [HCP-Users] Custom parcellation for HCP data Hi all, We want to create a custom parcellation for use with the connectome project data. However, we have not been able to achieve accurate spatial alignment between the random parcellation volume and the T1w volume. Specifically, there appears to be an "offset," possibly due to a mismatch between the origin of these volumes. In the following we describe the relevant steps of our pipeline. Please suggest any changes that might help to resolve the issue. 1. Create a high-resolution annotation (parcellation) by randomly parcellating the fsaverage surface into N regions of approximately equal volume. 2. Map the annotation from the source subject (fsaverage) to the target subject using mri_surf2surf. mri_surf2surf --srcsubject fsaverage --hemi lh --sval-annot highres.annot --trgsubject ${SUBJECTID} --srcsurfreg sphere.reg --trgsurfreg ${SUBJECTS_DIR}/${SUBJECTID}/MNINonLinear/Native/${SUBJECTID}.L.sphere.native.surf.gii --tval lh.highres.annot [repeat for rh] 3. Obtain vertices and faces data from ${SUBJECTS_DIR}/${SUBJECTID}/MNINonLinear/Native/${SUBJECTID}.L.white.native.surf.gii [repeat for rh] 4. Using the vertices and faces data obtained in step 3 as inputs for the write_surf Matlab function, create the lh.white surface file. [repeat for rh] 5. Obtain thickness data from ${SUBJECTS_DIR}/${SUBJECTID}/MNINonLinear/Native/${SUBJECTID}.L.thickness.native.shape.gii [repeat for rh] 6. Convert the annotation into a volume using mri_label2vol. mri_label2vol --annot lh.highres.annot --temp ${SUBJECTS_DIR}/${SUBJECTID}/T1w/T1w_acpc_dc_restore.nii.gz --identity --proj frac 0 1 .1 --subject ${SUBJECTID} --hemi lh --o vol_lh.nii [repeat for rh] 7. Configure the volume (remove unwanted ROIs). [repeat for rh] 8. Combine the configured volumes from each hemisphere to create the random parcellation volume. 9. Overlay the random parcellation volume on the template volume. See attached screenshot.jpeg showing the misalignment between the random parcellation volume and the template volume. Kind regards, Simon Baker Brain & Mental Health Laboratory Institute of Cognitive & Clinical Neuroscience Monash University _______________________________________________ HCP-Users mailing list [email protected]<mailto:[email protected]> 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 [email protected]<mailto:[email protected]> http://lists.humanconnectome.org/mailman/listinfo/hcp-users _______________________________________________ HCP-Users mailing list [email protected]<mailto:[email protected]> http://lists.humanconnectome.org/mailman/listinfo/hcp-users _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
