help
help > RE: Average BOLD Signal and GLM
Mar 19, 2020 01:03 PM | Alfonso Nieto-Castanon - Boston University
RE: Average BOLD Signal and GLM
Hi Nole,
Sorry, that information (average BOLD signal at each voxel) is not explicitly saved in CONN's pipeline (or at least not separately for each run/session; the file art_mean_*.nii contains the average BOLD signal value at each voxel across all runs/sessions). Typically the denoising step will remove the average BOLD signal within each voxel separately within each run/session, so your best approach is probably to take the fully-preprocessed but not yet denoised functional data (e.g. files named swau*.nii) and enter these in SPM first-level analyses in order to analyze the average BOLD signal values. If following that route, keep in mind that SPM first-level models are going to explicitly model the average BOLD signal within each run/session as a covariate of no interest as part of any first-level design matrix (these are typically named "Sn(###) constant" in in the SPM.xX.name variable), and also be aware of the default high-pass filtering (128s cutoff) step in SPM first-level analyses, since both of these procedures will attempt to remove the same effect that you are trying to characterize.
Hope this helps
Alfonso
Originally posted by Nole Hiebert:
Sorry, that information (average BOLD signal at each voxel) is not explicitly saved in CONN's pipeline (or at least not separately for each run/session; the file art_mean_*.nii contains the average BOLD signal value at each voxel across all runs/sessions). Typically the denoising step will remove the average BOLD signal within each voxel separately within each run/session, so your best approach is probably to take the fully-preprocessed but not yet denoised functional data (e.g. files named swau*.nii) and enter these in SPM first-level analyses in order to analyze the average BOLD signal values. If following that route, keep in mind that SPM first-level models are going to explicitly model the average BOLD signal within each run/session as a covariate of no interest as part of any first-level design matrix (these are typically named "Sn(###) constant" in in the SPM.xX.name variable), and also be aware of the default high-pass filtering (128s cutoff) step in SPM first-level analyses, since both of these procedures will attempt to remove the same effect that you are trying to characterize.
Hope this helps
Alfonso
Originally posted by Nole Hiebert:
I apologize if this has been answered elsewhere
but I was unable to locate an answer. I have preprocessed a
study containing four different, within-subject resting state
sessions in CONN and before I look at functional connectivity, I
want to compare average BOLD signal between sessions using
GLM. Is this possible to to within CONN? If not, are there
files I can export to look at in SPM?
Thanks!
Nole
Thanks!
Nole
Threaded View
| Title | Author | Date |
|---|---|---|
| Nole Hiebert | Mar 17, 2020 | |
| Alfonso Nieto-Castanon | Mar 19, 2020 | |