How to Map B0 and DWI Files to BIDS Format for DTI Data?

I used the bidsmapper function of bidscoin to find DTI-related files in my sourcedata folder. Bidscoin detected these three files:

  • B0diffPA64chDCCN16mm120b25001250
  • B0diffAP64chDCCN16mm120b25001250
  • diffPA64chDCCN16mm120b25001250

I found a README file in the sourcedata folder stating:

The two first files are the B0 images, often referred to as the “baseline” image, and are acquired with no diffusion weighting. B0 images are diffusion-weighted images acquired without diffusion weighting (i.e., b=0 s/mm²). They provide a reference for computing diffusion metrics (e.g., fractional anisotropy). All DWI images with higher b-values are compared to this unweighted image.The B0 images are based on the DWI acquisition parameters: they must match the distortions profile in the DWI data.

  • B0diffPA64chDCCN16mm120b25001250 is B0 PA encoding direction and should go into fmap folder

  • B0diffAP64chDCCN16mm120b25001250 is B0 AP encoding direction and should go into fmap folder

I’m trying to figure out which of these files are relevant and how they should be mapped into the BIDS output so that they can be digested by QSIPrep. Specifically:

  • Which ones go into the fmap/ folder?
  • Which go into the dwi/ folder?
  • What do I have to do to “link” those files together (I probably have to use the B0FieldSource or B0FieldIdentifier fields?)
  • I also have fieldmaps for my task fmri series so I guess I must make sure that fmriprep, etc. can differentiate between the files relevant for task fMRI and DTI?

P.S.: I am not entirely sure if this question is somehow related to this thread?

Hi @JohannesWiesner,

These will be your fmap/sub-X_dir-AP/PA_acq-DWI_epi.nii.gz.

QSIPrep only adopts IntendedFor at the moment.

Yes, that’s what IntendedFor and B0FieldSource/Identifier are for. That’s also why I added the acq-DWI to your filenames (purely aesthetic, but helps for human readers).

I would also make sure you have B0s in your main DWI sequence. If not, you might want to concatenate the according phase-encoding B0 into the main series. We can proceed that way if necessary.

Also, a small semantics note: DTI typically refers to low b-value low angular resolution (e.g., number of volumes) scans. Your file names suggest there are b-values as high as 2500, so I would not consider this DTI. Perhaps HARDI is better sequence name.

Best,
Steven

Thanks Steven!

I would also make sure you have B0s in your main DWI sequence. If not, you might want to concatenate the according phase-encoding B0 into the main series. We can proceed that way if necessary.

I am not sure if this was done (I am not the maintainer, just doing the BIDS-conversion here), but I will check. Is this a must?

Also, a small semantics note: DTI typically refers to low b-value low angular resolution (e.g., number of volumes) scans. Your file names suggest there are b-values as high as 2500, so I would not consider this DTI. Perhaps HARDI is better sequence name.

But QSIPrep would still work fine right? Because “HARDI” is not a valid BIDS entity?

Hi @JohannesWiesner

Yes

DTI is not a BIDS entity either. The suffix will be _dwi in any case.

Best,
Steven

So I check the resulting .bval files and they look like:

0 0 1250 2500 1250 2500 .... 0 1250 2500 1250 2500 1250 2500 ...

So that should be good right?

The suffix will be _dwi in any case.

Ah sorry, but that’s what I meant. Either way you would assing the _dwi sufffix to it :slight_smile:

Hi @JohannesWiesner,

Looks good!

Best,
Steven

1 Like