I have a question about the “_hcp_seechospacing” and “_hcp_bold_echospacing” parameters.
Which of these parameters is used when?
Looking at the HCP-pipeline scripts: I suspect that “_hcp_bold_echospacing” is only used during the hcp-prefreesurfer pipeline, and that only “–hcp_bold_echospacing” is used in the fmri-volume pipeline. And consequently, it seems to me that TOPUP (e.g.) therefore assumes that the fieldmaps have the same echospacing as given by “–hcp_bold_echospacing”.
Am I correct here?
Yes, --hcp_seechospacing is used in the hcp_pre_freesurfer command, it is mapped onto the seechospacing parameter of HCP PreFreeSurfer pipeline:
--hcp_seechospacing (str, default 'NONE'):
Echo Spacing or Dwelltime of Spin Echo Field Map or "NONE" if not
used.
On the other hand --hcp_bold_echospacing is used in the hcp_fmri_volume command and is mapped onto the echospacing parameter of the HCP fMRIVolume pipeline:
--hcp_bold_echospacing (float, default 0.00035):
Echo Spacing or Dwelltime of BOLD images.
You can get detailed technical documentation on Background — QuNex documentation or by adding --help to a QuNex command call. For example:
Thank you for further explaining! I was wondering why the parameter is specified for both resting state scans and fieldmaps, as I thought it should be equal anyways. But this might of course not be the case if fieldmaps are acquired that specifically match the structural acquisitions.
To extend a bit on that: I saw that the hcp1200 dataset actually always provides phase/magnitude fieldmap pairs for the structural scans and LR/RL pairs for the resting state acquisitions. However, in a qunex example (i.e. Session pipeline information files — QuNex documentation) the fieldmaps matching the functional data seem to be used for the structural data as well.
I am currently working with a dataset with fieldmaps that seem to be specifically matched to my resting state runs (same resolution and effective echotime). Would you know under what circumstances it is permissable to use these fieldmaps to correct for the distortions in the structural scans.
Yes, the two are often equal, but not always. Furthermore, there is also the hcp_dwi_echospacing for diffusion, which can again be different.
Sometimes se-fm pairs are acquired at the beginning and then used for structural + rest. For example, HCA (human connectome aging) acquired the se-fm pair, followed by structural + rest, followed by another se-fm pair and a set of tasks, and at the end another se-fm pair with another set of tasks. I am not an expert on acquisition protocols so I cannot really tell you what is “best” here, I have noticed many different things in the datasets I worked with.
I will ask my colleagues that are expert in this about the last point you made. Just so we are on the same page, your acquisition was T1w => T2w => SE-FM => rest BOLD? And you are wondering if you can use the SE-FM pair for distortion correction in structural data?
I just talked with a colleague that knows a lot more about MRI data acquisition than me. And if the assumption about your acquisition protocol T1w => T2w => SE-FM => rest BOLD is correct then you can very very likely use your SE-FM pair for distortion correction of structural data as well.
The reason is that magnetic field and distortions mostly remain relatively stable, at least over the time frame typical for such scans. Therefore, correcting the distortion on the structural images (T1w and T2w) using SE-FM, even if acquired later, would still be a reasonable practice. There is a slim chance of potential issues if shimming was used in an unconventional way (if shimming was done before T1w and T2w but then not again in front of SE-FM if I am not mistaken).