Fmriprep - T1 voxels not isotropic

Hi all,

I have T1 anatomical datasets with voxels that are not isotropic (130x512x512). I am converting the T1 anatomical datasets to BIDS using the raw dicoms. Once I get a nifti version of the T1 I can run it through flirt to isovoxel the brain (flirt -in t1brain -ref t1brain -applyisoxfm 1 -o t1brain_1mmiso.nii.gz). I would like to use the isovoxel T1 brain when running fmriprep.

I couldn’t find documentation indicating whether the T1 dataset is isovoxeled internal to fmriprep. If so, I can keep the T1 voxels in their native format and my concerns are addressed.

If not, does anyone know whether I would need to create a new json file for the isovoxeled T1 given that slice thickness and spacing between slices will change?

Any suggestions/pointers are welcome.

Thanks in advance,
FG

Out of curiosity, why?

No. fMRIPrep will not resize voxels unless you have two T1w images with mismatched dimensions. (Then it will take the smaller voxel size in each dimension.)

fMRIPrep does not use any BIDS metadata to guide T1w processing at this point. For the purposes of having a dataset with valid metadata, you’ll want to update that. For the purposes of fMRIPrep preprocessing, it won’t make a difference.

Hi Chris,

Thanks for the info RE metadata and fMRIPrep. I’ll definitely update the JSON files for posterity.

I’m working with clinical MRI data and will be drawing lesions on the T1 anatomical image. It’s difficult to accurately draw the lesions on a T1 with voxels that are vastly different in size.

Also I apply the T1-to-MNI space transformation matrix (output by fmriprep) to the lesion file to normalize the lesion to MNI space using ANTs (described here: Fmriprep – Failed normalisation with lesion mask - #12 by sandrushba).

It’s ideal to use the isovoxeled T1 brain in fmriprep in order to normalize the lesion. If I didn’t have to draw and normalize the lesion then I would leave the T1 voxels alone.

FG

1 Like

Hi Chris,

I wanted to follow up to this thread to ask if you know of a process by which to extract a json file with metadata from a nifti file.

Just to recap: I’ve isovoxeled my T1 anatomical data in order to draw lesions. The slice thickness and spacing between slices is now different.

Is it possible to create a new json with metadata from a nifti file or does one need to use the raw dicoms to get metadata? I’ve looked through various forums and have had difficulty finding a solution.

Thanks in advance for your help,
FG

Oh, I’m sorry for missing this.

I don’t know of any tools that will modify BIDS metadata from NIfTI header information, although I don’t think any header information would need to be synced here.

All good–we’ve found ways to modify the json file for long-term posterity. Thanks again for your input.