Feature Request / How to Implement: SIFT filtering for dMRI streamlines about connectomemapper3

[smeisler]

Great, thanks! I appreciate your quick responses to all of my questions, can't wait to get this program running on my end!

[sebastientourbier]

Hi Steven,

I have good news, this Nipype interface is already implemented in CMP3 here:

connectomemapper3/cmtklib/interfaces/mrtrix3.py

Lines 1334 to 1361 in 69d50b8

	class FilterTractogramInputSpec(CommandLineInputSpec):
	in_tracks = File(exists=True, mandatory=True, argstr='%s',
	position=-3, desc='Input track file')
	in_fod = File(exists=True, mandatory=True, argstr='%s', position=-2,
	desc='Input image containing the spherical harmonics of the fibre orientation distributions')
	out_file = File(argstr='%s', position=-1,
	desc='Output text file containing the weighting factor for each streamline')
	class FilterTractogramOutputSpec(TraitedSpec):
	out_weights = File(
	exists=True, desc='Output text file containing the weighting factor for each streamline')
	class FilterTractogram(MRTrix3Base):
	_cmd = 'tcksift2'
	input_spec = FilterTractogramInputSpec
	output_spec = FilterTractogramOutputSpec
	def _list_outputs(self):
	outputs = self.output_spec().get()
	if not isdefined(self.inputs.out_file):
	outputs['out_weights'] = op.abspath('streamlines_weights.txt')
	else:
	outputs['out_weights'] = op.abspath(self.inputs.out_file)
	return outputs

.
I guess the easiest way would be to add this Nipype interface node to MRtrix3 sift2 directly in the tracking subworkflow.
This means this would need the following changes:

• Updating imports from cmtklib.interfaces.mrtrix3 with FilterTractogram here:
  

  connectomemapper3/cmp/stages/diffusion/tracking.py

  Line 21 in 69d50b8

  from cmtklib.interfaces.mrtrix3 import Erode, StreamlineTrack

• Creating this new workflow node and rearranging the workflow connections here:
  

  connectomemapper3/cmp/stages/diffusion/tracking.py

  Lines 476 to 492 in 69d50b8

  	flow.connect([
  	(inputnode, mrtrix_tracking, [('DWI', 'in_file')]),
  	# (inputnode,mrtrix_tracking,[('wm_mask_resampled','mask_file')]),
  	# (mrtrix_tracking,outputnode,[('tracked','track_file')]),
  	# (mrtrix_tracking,converter,[('tracked','in_file')]),
  	# (mrtrix_tracking,converter,[('tracked','in_file')]),
  	# (inputnode,converter,[('wm_mask_resampled','image_file')]),
  	# # (converter,outputnode,[('out_file','track_file')])
  	# (converter,outputnode,[('out_tracks','track_file')])
  	# (mrtrix_tracking,converter,[('tracked','in_file')]),
  	# (inputnode,converter,[('wm_mask_resampled','image_file')]),
  	# (converter,outputnode,[('out_file','track_file')])
  	(mrtrix_tracking, converter, [('tracked', 'in_tracks')]),
  	(inputnode, converter, [('wm_mask_resampled', 'in_image')]),
  	(converter, outputnode, [('out_tracks', 'track_file')])
  	])

  
  It will look something like:

sift2 = pe.Node(interface=FilterTractogram(), name='sift2')
flow.connect([
            (inputnode, mrtrix_tracking, [('DWI', 'in_file')]),
            (mrtrix_tracking, sift2, [('tracked', 'in_tracks')]),
            (inputnode, sift2, [('DWI', 'in_fod')]),
            (sift2, converter, [('out_file', 'in_tracks')]),
            (inputnode, converter, [('wm_mask_resampled', 'in_image')]),
            (converter, outputnode, [('out_tracks', 'track_file')])
        ])

This will be integrated within the next release candidate I am currently working on (see PR #45).
However, I cannot tell you exactly a date but I guess you might have all recipe to integrate it on your side in the meantime.

[weimath]

Hi，
In the "ref_diffusion_config.json" file,
when set "mrtrix_tracking_config.sift": true,
This is use tcksift in MRtrix3, but I want use tcksift2, what should I do？
Please, thank you very much