search

ABCD-STUDY / nda-abcd-collection-3165

This is the documentation repository for ongoing releases to the NDA collection 3165. Please add GitHub Issues here for requests or feedback.
https://collection3165.readthedocs.io
BSD 3-Clause "New" or "Revised" License
16 stars 3 forks source link

Atlas space for dtseries #18

Closed seonjoo closed 3 years ago

seonjoo commented 3 years ago

For the minimally processed fMRI data, I can see

sub-_ses-baselineYear1Arm1_task-_bold_desc-filtered_timeseries.dtseries.nii

And it says that "Concatenated functional task dense time series post-DCANBOLDProcessing (regression and filtering) in Atlas space."

What does the Atlas space mean? Is it in MNI space? If I want to use Cole's Atlas (with subcortical parcellation in MNI space) is it correct way to extract parcels?

arueter1 commented 3 years ago

Hi There! Thanks for the question! We will get back to you with an answer as soon as possible.

ericfeczko commented 3 years ago

Hi Seonjoo!

My apologies for the late reply. The file is encoded in a CIFTI format, where the volumetric space is MNI, and the surface space is the fsaverage_LR32k ( an excellent "unofficial" explanation of the format can be found here, and see here for the associated manuscript on the associated formats). Briefly, the surface space is the freesurfer average template with the left and right hemispheres mirrored. Each hemisphere contains 32,000 vertices -- including the medial wall.

The workbench command cifti-separate can be used to split CIFTI files into GIFTI surface files and a volumetric subcortical file. The volumetric file for the dtseries here will contain the functional MRI timeseries data mapped to the FSL 2006 MNI template. These files will also be quite large.

An alternative is to use workbench to create a label file from the volumetric ROI. To do so, you'll need to check whether Cole's ROIs can be mapped to one of 19 subcortical structures indexed within the CIFTI. Such a label file can easily contain multiple labels. Afterwards, the -cifti-parcellate command can be used to extract a parcellated time-series per label within the label file.

Hope this helps!