This covers anatomical data prepared for the NSD dataset (e.g. T1 and T2 volumes) as well as FreeSurfer outputs.
This is the aseg.mgz (anatomical segmentation) file that is created by FreeSurfer but transformed (using winner-take-all) to the official NSD anatomical spaces and functional spaces. See FreeSurfer’s FreeSurferColorLUT.txt file (a copy is in nsddata/templates/FreeSurferColorLUT.txt) for interpretation of what the integer values mean. This information allows you to select white matter, CSF, ventricles, subcortical regions, etc.
subj01/anat/aseg_0pt8.nii.gz
The binary brain mask that was used to mask the anatomical volumes (e.g. T1, T2) for de-identification purposes. Note that this brain mask is intentionally liberal so as to not lose brain voxels.
subj01/anat/brainmask_0pt8.nii.gz
A version of the mean EPI volume that has been warped to the 1.0-mm anatomical space for that subject, as well as a version that has been warped to MNI space.
subj01/anat/EPI_to_anat1pt0.nii.gz
This is the automated FreeSurfer hippocampal segmentation that has been transformed (using winner-take-all) to the official NSD anatomical spaces and functional spaces.
subj01/anat/hippoSfLabels_0pt8.nii.gz
This shows locations of errors in FreeSurfer cortical surface reconstructions, as determined by visual inspection. There are generally few errors, and these errors occur in stereotypical locations (see NSD data paper).
subj01/anat/surfaceimperfections_0pt8.nii.gz
The official T1, T2, SWI, and TOF volumes for a given subject. These volumes have been masked. The different resolutions of the volumes all share the exact same field-of-view and exact same center.
subj01/anat/T1_0pt8_masked.nii.gz
Here, we took the pre-processed diffusion data, extracted the b0 volumes, averaged the b0 volumes within Run 1 and within Run 2, and then averaged the two averages together, producing a single volume (at 0.8-mm resolution). This volume was then resampled to different resolutions (in the same manner as the other anatomical volumes).
Versions of the volumes that have been warped to MNI space.
subj01/anat/T1_to_MNI.nii.gz
This is the final FreeSurfer directory for subject AA, reflecting the result of manual edits to the tissue segmentation.
In running FreeSurfer, a 0.8-mm T1 volume was provided to FreeSurfer and the ‘-hires’ flag was used. Also, we have performed additional FreeSurfer-related processing, which created additional files. The changes include (1) creating layerB1, layerB2, and layerB3 surfaces which correspond to 25%, 50%, and 75% of the distance from the pial surface to the white-matter surface; (2) creating semi-inflated surfaces (e.g. ?h.semiinflated + ?h.sulcsemiinflated); and (3) creating flattened cortical surfaces (e.g. ?h.full.flat.patch.3d)). Also, note that the manually edited subject directory has modified files: for example, the brainmask.mgz file has had “holes” put into it (to aid in the surface reconstruction process).
Note that FreeSurfer has a built-in fsaverage flattened surface called [lh,rh].cortex.patch.flat. This is distinct from the flattened cortical surfaces described above. Note that the two flattened surfaces are rotated differently, so one may need to rotate the surfaces to a more canonical orientation for visualization purposes. Also, note that the full-cortex flattened surfaces remove substantial cortex near the midline (e.g. cingulate cortex), so be careful when interpreting results.
Besides the typical FreeSurfer outputs, the subject directories also contain a number of NSD-specific data files. These include ROI files and results from the prf, floc, and NSD experiments.
The FreeSurfer special "fsaverage" subject. Again, additional files are present in this directory, reflecting additional FreeSurfer-related processing that we have performed.
These are standard FreeSurfer directories. No additional files are present in these directories.
This is the original, non-edited FreeSurfer output for subject AA. Note that the surfaces in the edited and original versions are not compatible with one another, given that they have different numbers of vertices.
This the raw FreeSurfer output produced when run on an individual T1 acquisition (the BBth one) for subject AA. (Please note that the individual T1 acquisitions processed here are the co-registration procedure; hence, all of the results should be directly spatially comparable.) The call to FreeSurfer was the same as the original FreeSurfer call, except that the -hippocampal-subfields option was run with -T1 not -T1T2. Furthermore, no additional FreeSurfer-related processing was run for these directories.
These individual T1 FreeSurfer directories may be useful for assessing the reliability of FreeSurfer outputs for individual subjects. However, note that the final FreeSurfer directory (freesurfer/subjAA) reflects manual edits to the segmentation. Thus, a more appropriate comparison may be to use the freesurferoriginals/subjAA_original directory.