Fmriprep warning: No single-band-reference found for

Following warning is shown when running preprocessing:
"No single-band-reference found for "


If you don’t have a single-band reference, then this is just a notification. If you do, could you share the tree output for a subject?

1 Like

We don’t have- I just wanted to know if it’s likely to hurt our results in any significant way.
Thanks a lot!

If you are using a multiband sequence that results in images with low gray/white tissue contrast, acquiring a single-band reference immediately prior to your multiband sequence can assist in any steps that benefit from high contrast, such as BOLD-T1w registration and some susceptibility distortion correction methods (the SyN-based, fieldmap-less correction, for instance).

1 Like

Thanks for your kind answer!
I have a few more questions on this topic:

  1. If my data uses a multi-band(MB) acquisition, then without a single-band reference, will fMRIPrep automatically do slice-timing-correction adapted to MB acquisition?
  2. How can I design the experiment to acquire a single-band reference? In my oppinion, each run should have such a reference, then how many TR should I use for collecting this reference? And what reqirements should be met to make fMRIPrep handle suh situation? If there is any related resources, please let me know, Thanks!

If you have SliceTiming metadata defined, then fMRIPrep will automatically correct with AFNI’s 3dTshift unless you use the --ignore slicetiming option.

As I understand it, the usual way to acquire this is one single-band EPI acquisition immediately prior to each BOLD run. Since it’s a single EPI volume, and you don’t care about non-steady-state T1 relaxation effects, it should take no more than 3s to acquire, probably less. You’ll want it to have the same voxel size and field of view as your MB sequence.

I also got the following warning:
"No single-band-reference found for ". I do have a single-band-reference file acquired. Shall I place in the func folder for fmriprep to recognise it? Also what will be its BIDS nomenclature?



Depending on which MRI sequence you are using to produce your multi-band functional images, you may get a single single-band reference (sbref) image for each functional run.
I know that is the way it works with the CMRR sequence on Siemens scanner. For fmriprep to see these sbref images, the sbref images should be placed in the func directory with the same name as the associated functional run, except that the _bold suffix should be replaced by _sbref.

Regarding the time used to acquire this reference, again, for the CMRR sequence, it takes as many TRs as the multi band acceleration factor. For example, if you have a TR of 1s, and a multiband acceleration factor of 3, then you would get 3s of scan used at the beginning of the acquisition to acquire the sbref image, then X TR to acquire the dummy scans, and then the acquisition would proceed with the acquisition of the functional runs themselves.