[RESOLVED] Best practices for combining pairs of reverse phase encoded resting scans (HCP-style acquisition)

1

Hello Qunex’ers,

I have searched the forums and the literature but I remain unsure of best practices for combining reverse PE scans within the qunex/HCP processing framework. We collected data in an HCP-style acquisition and we are using qunex to analyze the data; our resting sequences are as follows:

rfMRI_REST_AP
rfMRI_REST_PA
rfMRI_REST_AP
rfMRI_REST_PA

My questions are:

  1. Exactly how are multiple scans like these intended to be combined in order to leverage the benefits of having collected them in reverse PE pairs? Should I use wb_command (i.e. cifti-merge) to simply combine across all 4 (demeaned) scans? My intuition says this may not be quite right – is there a more ideal way to combine the 2 pairs and then subsequently use -cifti-merge?
  2. At what stage in the pipeline should the scans be averaged/combined? Based on what I see in the minimally processed data from HCP, it seems like this is done at some point after the minimal processing pipeline–at what stage of the qunex pipeline should I combine the data? E.g., if I want to look at GBC of the combined rFC data per subject, should I merge the 4 scans after preprocess_bold or after compute_bold_fc

Thanks in advance for any insights. If this has been answered before or is included with some HCP/qunex documentation that I missed, I do apologize and I would be happy to be pointed in the right direction.

My very best,
Rachael

2

Hi Rachael,

While I try to get someone that has more knowledge I can provide my very limited input on the subject. You onboard the scans separately into QuNex, when you run them through HCP minimal preprocessing pipeline (hcp_pre_freesurfer, hcp_freesurfer, hcp_post_freesurfer, hcp_fmri_volume, hcp_fmri_surface) they will be processed independently.

When you perform highly recommended additional HCP denoising (hcp_icafix, hcp_msmall) the processing will use all of them simultaneously. Here, one of the outputs is also a cleaned up and denoised concatenated (in your case concatenation of 4 resting state BOLDs) BOLD image. On our end, we commonly continue with analyses (e.g., connectome, activation) on that image.

Jure

3

Thank you for this, Jure. I will explore the functionality of hcp_icafix and hcp_msmall while I wait for further insight.

4

Hi Rachael,

I talked with my colleagues and they said when performing some bespoke analyses outside of QuNex, they usually either manually concatenate BOLDs at the end of preprocessing or use the concatenated MSMAll output.

If you use QuNex analysis commands (e.g., for activation analyses or calculating connectomes) this concatenation will be executed automatically through conc or list files.

Jure

5

Thanks, Jure, really appreciate the additional insight. I’ll start with hcp_msmall and see how it goes.

6

Please let me know if you need any assistance with hcp_icafix or hcp_msmall.