-
Notifications
You must be signed in to change notification settings - Fork 532
/
Copy pathdmri_group_connectivity_mrtrix.py
180 lines (132 loc) · 7.33 KB
/
dmri_group_connectivity_mrtrix.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
"""
==================================================
dMRI: Group connectivity - MRtrix, FSL, FreeSurfer
==================================================
Introduction
============
This script, dmri_group_connectivity_mrtrix.py, runs group-based connectivity analysis using
the dmri.mrtrix.connectivity_mapping Nipype workflow. Further detail on the processing can be
found in :doc:`dmri_connectivity_advanced`. This tutorial can be run using::
python dmri_group_connectivity_mrtrix.py
We perform this analysis using one healthy subject and two subjects who suffer from Parkinson's disease.
The whole package (960 mb as .tar.gz / 1.3 gb uncompressed) including the Freesurfer directories for these subjects, can be acquired from here:
* http://db.tt/b6F1t0QV
A data package containing the outputs of this pipeline can be obtained from here:
* http://db.tt/elmMnIt1
Along with MRtrix, FSL, and Freesurfer, you must also have the Connectome File Format
library installed as well as the Connectome Mapper (cmp).
* MRtrix: http://www.brain.org.au/software/mrtrix/
* FSL: http://www.fmrib.ox.ac.uk/fsl/
* Freesurfer: http://surfer.nmr.mgh.harvard.edu/
* CTMK: http://www.cmtk.org/
* CFF: sudo apt-get install python-cfflib
Or on github at:
* CFFlib: https://github.com/LTS5/cfflib
* CMP: https://github.com/LTS5/cmp
Output data can be visualized in ConnectomeViewer, TrackVis, Gephi,
the MRtrix Viewer (mrview), and anything that can view Nifti files.
* ConnectomeViewer: https://github.com/LTS5/connectomeviewer
* TrackVis: http://trackvis.org/
* Gephi: http://gephi.org/
The fiber data is available in Numpy arrays, and the connectivity matrix
is also produced as a MATLAB matrix.
Import the workflows
--------------------
First, we import the necessary modules from nipype.
"""
import nipype.interfaces.fsl as fsl
import nipype.interfaces.freesurfer as fs # freesurfer
import os.path as op # system functions
import cmp
from nipype.workflows.dmri.mrtrix.group_connectivity import create_group_connectivity_pipeline
from nipype.workflows.dmri.connectivity.group_connectivity import (create_merge_network_results_by_group_workflow, create_merge_group_network_results_workflow, create_average_networks_by_group_workflow)
"""
Set the proper directories
--------------------------
First, we import the necessary modules from nipype.
"""
subjects_dir = op.abspath('groupcondatapackage/subjects/')
data_dir = op.abspath('groupcondatapackage/data/')
fs.FSCommand.set_default_subjects_dir(subjects_dir)
fsl.FSLCommand.set_default_output_type('NIFTI')
"""
Define the groups
-----------------
Here we define the groups for this study. We would like to search for differences between the healthy subject and the two
vegetative patients. The group list is defined as a Python dictionary (see http://docs.python.org/tutorial/datastructures.html),
with group IDs ('controls', 'parkinsons') as keys, and subject/patient names as values. We set the main output directory as 'groupcon'.
"""
group_list = {}
group_list['controls'] = ['cont17']
group_list['parkinsons'] = ['pat10', 'pat20']
"""
The output directory must be named as well.
"""
global output_dir
output_dir = op.abspath('dmri_group_connectivity_mrtrix')
"""
Main processing loop
====================
The title for the final grouped-network connectome file is dependent on the group names. The resulting file for this example
is 'parkinsons-controls.cff'. The following code implements the format a-b-c-...x.cff for an arbitary number of groups.
.. warning::
The 'info' dictionary below is used to define the input files. In this case, the diffusion weighted image contains the string 'dti'.
The same applies to the b-values and b-vector files, and this must be changed to fit your naming scheme.
The workflow is created given the information input about the groups and subjects.
.. seealso::
* nipype/workflows/dmri/mrtrix/group_connectivity.py
* nipype/workflows/dmri/mrtrix/connectivity_mapping.py
* :doc:`dmri_connectivity_advanced`
We set values for absolute threshold used on the fractional anisotropy map. This is done
in order to identify single-fiber voxels. In brains with more damage, however, it may be necessary
to reduce the threshold, since their brains are have lower average fractional anisotropy values.
We invert the b-vectors in the encoding file, and set the maximum harmonic order
of the pre-tractography spherical deconvolution step. This is done to show
how to set inputs that will affect both groups.
Next we create and run the second-level pipeline. The purpose of this workflow is simple:
It is used to merge each subject's CFF file into one, so that there is a single file containing
all of the networks for each group. This can be useful for performing Network Brain Statistics
using the NBS plugin in ConnectomeViewer.
.. seealso::
http://www.connectomeviewer.org/documentation/users/tutorials/tut_nbs.html
"""
title = ''
for idx, group_id in enumerate(group_list.keys()):
title += group_id
if not idx == len(list(group_list.keys())) - 1:
title += '-'
info = dict(dwi=[['subject_id', 'dti']],
bvecs=[['subject_id', 'bvecs']],
bvals=[['subject_id', 'bvals']])
l1pipeline = create_group_connectivity_pipeline(group_list, group_id, data_dir, subjects_dir, output_dir, info)
# Here with invert the b-vectors in the Y direction and set the maximum harmonic order of the
# spherical deconvolution step
l1pipeline.inputs.connectivity.mapping.fsl2mrtrix.invert_y = True
l1pipeline.inputs.connectivity.mapping.csdeconv.maximum_harmonic_order = 6
# Here we define the parcellation scheme and the number of tracks to produce
parcellation_name = 'scale500'
l1pipeline.inputs.connectivity.mapping.Parcellate.parcellation_name = parcellation_name
cmp_config = cmp.configuration.PipelineConfiguration()
cmp_config.parcellation_scheme = "Lausanne2008"
l1pipeline.inputs.connectivity.mapping.inputnode_within.resolution_network_file = cmp_config._get_lausanne_parcellation('Lausanne2008')[parcellation_name]['node_information_graphml']
l1pipeline.inputs.connectivity.mapping.probCSDstreamtrack.desired_number_of_tracks = 100000
l1pipeline.run()
l1pipeline.write_graph(format='eps', graph2use='flat')
# The second-level pipeline is created here
l2pipeline = create_merge_network_results_by_group_workflow(group_list, group_id, data_dir, subjects_dir, output_dir)
l2pipeline.inputs.l2inputnode.network_file = cmp_config._get_lausanne_parcellation('Lausanne2008')[parcellation_name]['node_information_graphml']
l2pipeline.run()
l2pipeline.write_graph(format='eps', graph2use='flat')
"""
Now that the for loop is complete there are two grouped CFF files each containing the appropriate subjects.
It is also convenient to have every subject in a single CFF file, so that is what the third-level pipeline does.
"""
l3pipeline = create_merge_group_network_results_workflow(group_list, data_dir, subjects_dir, output_dir, title)
l3pipeline.run()
l3pipeline.write_graph(format='eps', graph2use='flat')
"""
The fourth and final workflow averages the networks and saves them in another CFF file
"""
l4pipeline = create_average_networks_by_group_workflow(group_list, data_dir, subjects_dir, output_dir, title)
l4pipeline.run()
l4pipeline.write_graph(format='eps', graph2use='flat')