Hi,
I have started learning about neuroscience and encountered a question that I hope someone can help me with. Thank you very much in advance.
When processing BOLD data with XCPD, adding the --cifti flag results in the output file sub-HC001_task-rest_acq-ap_run-1_space-fsLR_den-91k_desc-denoisedSmoothed_bold.dtseries.nii, while not setting the --cifti flag (with all other parameters remaining the same) generates the file sub-HC001_task-rest_acq-ap_run-1_space-MNI152NLin6Asym_res-2_desc-denoisedSmoothed_bold.nii.gz. I understand that one is based on vertex-based computation and the other on voxel-based computation. However, both files should contain subcortical information. Are the subcortical data contained in the two output datasets numerically equal? Or, to put it another way, if I use the same brain atlas (for example: Brainnetome 246) to extract subcortical time series from the .dtseries.nii file and from the .nii.gz file, will the extracted values be the same?
The normalization to MNI152NLin6Asym or fsLR is handled by the preprocessing pipeline, not XCP-D, so this isn’t really an XCP-D question. If the input data are numerically equal (i.e., the preprocessing pipeline normalized the subcortical data in the CIFTI and the whole brain in the NIfTI in exactly the same way) then XCP-D should produce essentially the same results. There will almost certainly be small differences between extracted time series from the CIFTI and NIfTI, but I can’t say whether those come from floating point errors, differences in the two versions of the atlas, differences in preprocessing, or differences in how XCP-D handles the data.