Dcm2bids with multi-echo BOLD data

Hi Neurostars’ers,

Thanks in advance for reading this!

We’re using fmriprep_23.1.4. We have collected a BOLD fMRI dataset with multiple-echos (which we intend to analyze with fMRIprep and TEDANA). The dataset includes a pair of short AP-PA acquisitions for calculating fieldmaps. Each scan also has a single-band reference image. Based on dcm2bids_helper, this results in the following nifti files:

1a. Short AP sbref
1b. Short AP echo 1
1c. Short AP echo 2
1d. Short AP echo 3

2a. Short PA sbref
2b. Short PA echo 1
2c. Short PA echo 2
2d. Short PA echo 3

3a. Long PA sbref
3b. Long PA echo 1
3c. Long PA echo 2
3d. Long PA echo 3

I was wondering what the best approach is for running dcm2bids on this dataset. Specifically:

  1. Should the fieldmaps be based only on the shortest echo time, which has the best contrast? Or is there a way to define a fieldmap based on the optimal sum of echos?
  2. Should we simply ignore the sbref scans for the short scans that we collected for fieldmapping?

Many thanks!

You should have an SBRef for each echo - and I would use the shortest echo SBRef for fieldmap related parts of processing. I don’t think an optimal combination would help you, as you say the contrast is good.

Thank you for your advice, @dowdlelt!
And you’re absolutely right. We do have an SBRef per echo.

(For future reference, a similar quesiton had been discussed here before)