Fmriprep without bids validation

Dear all,

Can i run fmriprep without BIDS validation ?, my folder structure format is not in BIDS and i dont want to use currently BIDS format, can i possibly run fmriprep preprocessing in such situation ?. Could you please guide me how i can do that

Thanks
Vasudev

Hi @Vaasudev. fMRIPrep assumes a BIDS dataset. We can’t make any assurances of behavior if your data do not conform. --skip-bids-validation exists for situations where the conformance issues are peripheral, and all of the necessary data (T1w, T2w, fieldmaps, bold) is in BIDS format. But if your folder structure is not BIDS, it is very likely not to work with fMRIPrep.

1 Like

Is there any command line automation script to transform my nifti data into BIDS, i do not have DICOM data for my subjects at the moment.

I have seen automated dcm2bids converter, it is pretty bad.

Is there any alternative, if so please kindly let me know

Thanks
Vasudev

You could have a look at bidsify. There’s a list of other converters on the website, as well.

1 Like

My real problem is to even automate the creation of config files, yes bidsify is requires one to create a config file, but i simply want a procedure where i can specify the path the subject folder directory where T1w,fMRI,T2w and DWI dicoms are situated and the script automatically creates bids directories for corresponding dicom files.

I am trying to prepare a protocol that medical team in my hospital can use my tools with minimal intervention from their end. I can indeed use dcm2nii even but i have to create .json files, which i can do it but i want to automate that procedure competely.

Is there a such automated pipeline?

No. There isn’t a general solution to take a directory full of idiosyncratically named files and produce a structured output.

If you can talk with the MRI techs, you could get them to configure the scanner to output DICOMs named according to ReproIn, which can be converted BIDS with a single call to heudiconv. But that will only help with newly collected datasets.