MRIQC custom brain template

I want to use MRIQC to extract metrics for ultra-low-field, infant MRIs. Because of the differences in brain shape and in image contrast, I would like to use our own (age-specific) templates, and I have a couple of questions:

  1. What files are required to be included with the template? Feeding the mriqc repo into an AI tool (Claude), it reports that it only needs:
  • res-02_desc-brain_mask — normalization reference mask
  • res-01_label-CSF_probseg, res-01_label-GM_probseg, res-01_label-WM_probseg — probability masks
  • res-02_desc-fMRIPrep_boldref
  • res-01_desc-carpet_dseg

Is this accurate?

  1. If only those files are actually needed, would it be OK to remove the other hundred files included in the nipreps/mriqc Docker image, under “/templateflow/tpl-MNI152NLin2009cAsym” to reduce the size of the image?

Is this accurate?

Sounds about right

If only those files are actually needed, would it be OK to remove the other hundred files included in the nipreps/mriqc Docker image, under “/templateflow/tpl-MNI152NLin2009cAsym” to reduce the size of the image?

Yes, absolutely. Please send a PR removing anything that’s not needed!

Regarding your initial question on how to feed custom templates, please check Python Client - TemplateFlow . Adding your custom template to templateflow would then allow MRIQC to select it for processing.

We’d be very interested in adding your new features to MRIQC, should you send a PR!