Thank you for the help! To clarify, by "native space" I had meant "without spatial normalization to MNI space". But on reflection that's probably not the proper way to think about or describe this data. As I understand it, when we download the preprocessed functional data we get two versions of the same information: volumetric (e.g. tfMRI_MOTOR_LR.nii.gz) and surface (by grayordinates, e.g. tfMRI_MOTOR_LR_Atlas.dtseries.nii).
These correspond for timepoints: the 4th dimension in the .nii.gz has 284 images, the dtseries.nii time course chart (in the Workbench) has 284 points along the x-axis, and the text file the code you provided creates has 284 lines. But, there won't be 1-to-1 mapping between timeseries in the .nii.gz and dtseries.nii, because the first is summarized into voxels and the second is surface-based. To my next question, is it possible to extract not the timeseries averaged over all ROI grayordinates (-cifti-roi-average), but for each individual grayordinate specified in the ROI mask? If so, how is the adjacency specified? (e.g grayordinate 1 shares a side with grayordinates 2, 3, and 4). For context, I wish to perform a ROI or searchlight-type MVPA on data from the HCP. To do this with 4d volumetric images I first get 3d masks of the voxels to include, then extract the timeseries for those voxels (e.g. as a text matrix), and can use the coordinates to get adjacency information. While I already can use the .nii.gz images for this analysis, I would like to perform it surface-wise as well. thanks again, Jo _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users
