Use frmriprep fieldmap-less function with 1.5T multiband data?

Software Support
1

Hopefully this is the right place for this question - redirect me if its not.

I’m analysing the data from the NNDb dataset, OpenNeuro, using fmriprep for preprocessing, and hoping to get the data into both volume and surface spaces. The dataset does not contain fieldmaps. The question I have is whether I should use the fieldmap-less option.

The data is movie watching, ~whole brain coverage (TR=1, 3.2mm isotropic), from a 1.5T scanner, 4x multiband, no in-plane acceleration. Arguments against - distortion is minimal with 1.5T, multiband images may not be optimal for the fieldmap less algorithm. Arguments for - all the normal ones. I have tried running a sample of brains through, in most cases the fieldmap-less algorithm seems to do a good job, in others it doesn’t (e.g. stretched occipital lobe).

Any advice/thoughts? [ References would of course be really helpful as well so I can justify any decision. ]

2

I’ve answered my own question empirically -

I ran the whole dataset through with --use-syn-sdc. I did the same with a more standard 3T dataset, OpenNeuro. By eye I can see no obvious difference in the quality of the distortion corrections. So I guess it works.