Proper imaging parameters for PEPOLAR spatial distortion correction in fMRIPrep

Software Support
1

Hello experts,

The HCP appendix lists these requirements for the SE-EPI protocol:

You should then create a SE-EPI protocol matched in resolution, FOV, matrix size, bandwidth, and echo spacing to what you’ll use for your fMRI acquisitions (i.e., your gradient-echo EPI protocol). TR/TE doesn’t have to match.

fmriprep uses topup for spatial distortion correction. My question is in regards to the effective echo spacing. Is matching the effective echo spacing between the SE and fMRI acquisitions a requirement for proper spatial distortion correction in fmriprep?

2

I don’t know much about the physics myself, but in our own imaging protocols and a few of the other sites we process data for, the effective echo spacing is the same between the SE and fMRI.

Are you designing a protocol or is this a retrospective study?

3

Both designing and retrospective. Some of our past studies did not match the effective echo spacing between the SE and fMRI; there was a difference of 8 microseconds. With these studies, I notice the spatial distortion correction is a bit off with fmriprep versions using topup for the correction. With older fmriprep versions using 3dQwarp, the spatial distortion correction looks much better. For all our future studies, I am trying to match the effective echo spacing exactly between the SE and fMRI.

4

In one of our older protocols (based off of ABCD sequences), the effective echo spacing between our SE-EPI and fMRI is ~ 4.8 microseconds. When I tested fMRIprep (~v21.0.0), TOPUP could not correct the fMRI.

If you need to rescue your retrospective data and TOPUP is not doing a good job, you may need to test a fieldmapless technique like SyN-SDC or SynBOLD-DISCO.

1 Like
5

That is unfortunate.

I was hoping that was not the case. The second post here SDC in fmriprep guessed “sub-millisecond changes in TRT are going to have pretty minimal effect.” It doesn’t reference effective echo spacing, but I believe TRT and EES are mathematically related somehow. The third post references the HCP documentation on which parameters should match between fMRI and spin-echo sequences and echo spacing is one of those parameters. I have been using the documentation for setting fMRI and SE parameters in a protocol for a future study.