Identify sub-workflows that are simpler and port them first

[yibeichan]

Spatial Normalization to MNI (ANTs) requires:

• niworkflows.interfaces.reportlets.registration.SpatialNormalizationRPT
  • python, integrate ants. Registration interface (nipype)
• templateflow.api.get

Calculate air mask above the nasial-cerebelum plane – airmsk_wf() requires:

• mriqc.interfaces.RotationMask
  • python, scipy.ndimage
• nipype.interfaces.ants.ApplyTransforms
• templateflow.api.get
• mriqc.interfaces.RotationMask
  • python, scipy.ndimage

Brain tissue segmentation (FAST) requires:

• nipype.interfaces.fsl.FAST

Extraction of IQMs – compute_iqms() requires

• niworkflows.interfaces.bids.ReadSidecarJSON
• mriqc.interfaces.reports.AddProvenance
  • nipype.utils.filemanip.hash_infile
• niworkflows.interfaces.anatomical.Harmonize
• mriqc.interfaces.ComputeQI2
  • python, numpy
• mriqc.interfaces.StructuralQC
  • python, numpy
• nipype.interfaces.ants.ApplyTransforms
• templateflow.api.get
• niworkflows.interfaces.anatomical.IQMFileSink
• niworkflows.utils._tofloat

[sahahn]

HeadMaskWorkflow - mriqc.workflows.anatomical.headmsk_wf

Either uses: fsl.BET(surfaces=True) which is already implemented, I think - (https://github.com/nipype/pydra-fsl/blob/master/pydra/tasks/fsl/preprocess/bet.py)

but if dipy is installed, then it uses a different workflow, which seems mostly based on python / nibabel functions, and then needs: nipype.interfaces.dipy import Denoise which would need to be implemented I believe as I do not see an existing interface for dipy, but dipy is a python library so might not be difficult

[mckenziephagen]

Skull stripping requires: nirodents.workflows.brainextraction.init_rodent_brain_extraction_wf and niworkflows.anat.skullstrip.afni_wf which requires mriqc.interfaces.synthstrip.SynthStrip niworkflows.interfaces.nibabel.IntensityClip niworkflows.interfaces.nibabel.ApplyMask nipype.interfaces.ants.N4BiasFieldCorrection

[garciaml]

Conform (reorientations, revise data types) input data and read associated metadata.

mriqc.workflows.anatomical.anat_qc_workflow

_# 1. Reorient anatomical image to_ras = pe.Node(ConformImage(checkdtype=False), name="conform")

Next it is used as a first node in the workflow (still in mriqc.workflows.anatomical.anat_qc_workflow)

_# Connect all nodes

fmt: off

workflow.connect([
    (inputnode, to_ras, [("in_file", "in_file")]), ..._

ConformImage defined in mriqc.interfaces.common.conform_image

It uses os.path, nibabel, mriqc.config, mriqc.messages, mriqc.interfaces.data_types, and from nipype.interfaces.base : BaseInterfaceInputSpec, File, SimpleInterface, TraitedSpec, traits

[oesteban]

Closing in favor of https://github.com/nipreps/mriqc/discussions/1347