Realistic Pipeline Generation Example

As a realistic pipeline generation usage example we will create a basic brain MRI preprocessing pipeline using Nipype’s interface for FSL.

The pipeline will receive a NIfTI instance from the database as input and then run:

1. Brain extraction (BET)
2. Reorientation to the MNI152 standard brain template space (fslreorient2std)
3. Cropping (robustfov)
4. Linear registration (FLIRT)
5. Nonlinear registration (FNIRT)

django_mri already provides the required analyses, so in order to create a new pipeline we can simply use a dict defining the pipes that make up the pipeline.

First, this pipeline assumes the MNI152 standard brain template exists in the database:

basic_fsl_preprocessing.py

from django_mri.models.nifti import NIfTI


try:
    MNI = NIfTI.objects.get(path__contains="MNI152_T1_2mm_brain")
except NIfTI.DoesNotExist:
    raise NIfTI.DoesNotExist("Could not find MNI152_T1_2mm_brain in the database.")

Then, in order to keep things readable, let’s define each node’s configuration and create a definition of that node:

BET_CONFIGURATION = {"robust": True}
REORIENT_CONFIGURATION = {}
ROBUSTFOV_CONFIGURATION = {}
FLIRT_CONFIGURATION = {"reference": MNI.id, "interp": "spline"}
FNIRT_CONFIGURATION = {"ref_file": MNI.id}

BET_NODE = {"analysis_version": "BET", "configuration": BET_CONFIGURATION}
REORIENT_NODE = {
    "analysis_version": "fslreorient2std",
    "configuration": REORIENT_CONFIGURATION,
}
ROBUST_FOV_NODE = {
    "analysis_version": "robustfov",
    "configuration": ROBUSTFOV_CONFIGURATION,
}
FLIRT_NODE = {
    "analysis_version": "FLIRT",
    "configuration": {"reference": MNI.id, "interp": "spline"},
}
FNIRT_NODE = {
    "analysis_version": "FNIRT",
    "configuration": {"ref_file": MNI.id},
}

Now we can specify the pipes:

BET_TO_REORIENT = {
    "source": BET_NODE,
    "source_port": "out_file",
    "destination": REORIENT_NODE,
    "destination_port": "in_file",
}
REORIENT_TO_FOV = {
    "source": REORIENT_NODE,
    "source_port": "out_file",
    "destination": ROBUST_FOV_NODE,
    "destination_port": "in_file",
}
FOV_TO_FLIRT = {
    "source": ROBUST_FOV_NODE,
    "source_port": "out_roi",
    "destination": FLIRT_NODE,
    "destination_port": "in_file",
}
FLIRT_TO_FNIRT = {
    "source": FLIRT_NODE,
    "source_port": "out_file",
    "destination": FNIRT_NODE,
    "destination_port": "in_file",
}

And finally, everything is ready for the pipeline:

BASIC_FSL_PREPROCESSING = {
    "title": "Basic FSL Preprocessing",
    "description": "Basic MRI preprocessing pipeline using FSL.",
    "pipes": [BET_TO_REORIENT, REORIENT_TO_FOV, FOV_TO_FLIRT, FLIRT_TO_FNIRT],
}

Add the pipeline to the database:

>>> from basic_fsl_preprocessing import BASIC_FSL_PREPROCESSING
>>> from django_analyses.models import Pipeline
>>> from django_analyses.pipeline_runner import PipelineRunner
>>> from django_mri.models import Scan

>>> pipeline = Pipeline.objects.from_dict(BASIC_FSL_PREPROCESSING)
>>> scan = Scan.objects.filter(description__icontains="MPRAGE").first()
>>> pipeline_input = {"in_file": scan.nifti}
>>> pipeline_runner = PipelineRunner(pipeline)
>>> results = pipeline_runner.run(inputs=pipeline_input)
>>> results
{<Node:
Node #51
FLIRT v6.0.3:b862cdd5
Configuration: [{'interp': 'spline', 'reference': 585}]
>: <Run: #117 FLIRT v6.0.3:b862cdd5 run from 2020-06-01 12:34:51.103207>,
<Node:
Node #49
BET v6.0.3:b862cdd5
Configuration: [{'robust': True}]
>: <Run: #114 BET v6.0.3:b862cdd5 run from 2020-06-01 12:34:39.941216>,
<Node:
Node #44
FNIRT v6.0.3:b862cdd5
Configuration: [{'ref_file': 585}]
>: <Run: #118 FNIRT v6.0.3:b862cdd5 run from 2020-06-01 12:35:28.673283>,
<Node:
Node #43
robustfov v6.0.3:b862cdd5
Configuration: [{}]
>: <Run: #116 robustfov v6.0.3:b862cdd5 run from 2020-06-01 12:34:48.512874>,
<Node:
Node #42
fslreorient2std v6.0.3:b862cdd5
Configuration: [{}]
>: <Run: #115 fslreorient2std v6.0.3:b862cdd5 run from 2020-06-01 12:34:46.985302>}

To get our output file we could:

>>> from django_analyses.models import Run
>>> from django_mri.models import NIfTI
>>> run = Run.objects.get(id=118)
>>> for output in run.output_set.all():
>>>     print(output.key, output.value)
fieldcoeff_file NIfTI object (653)
log_file /media/dir/analysis/118/log.txt
modulatedref_file NIfTI object (652)
warped_file NIfTI object (649)
field_file NIfTI object (650)
jacobian_file NIfTI object (651)
>>> path = NIfTI.objects.get(id=649).path
>>> path
/media/dir/analysis/118/warped.nii.gz