Dwi_whole_brain_tractograpy pipeline¶

Brain Tractography with MRTrix¶

The aim of this pipeline is to create a whole-brain tractogram for a multishell DWI using Constrained Spherical Deconvolution (CSD). In this pipeline, an image with b=0 volumes with opposite phase encoding should be provided for the purpose of EPI distortion correction. An anatomical image is also required.

This pipeline used several algorithms in order to improve the biological plausibility of fiber tracking :
- Anatomically Constrained Tractography (ACT) which rejects streamlines that end in biologically implausible tissue
- Spherical-deconvolution informed filtering of tractograms (SIFT) which corrects for the fact that longer streamlines tend to be overestimated in tractography
- Multi-shell multi-tissue (MSMTT) CSD which exploites the differences in b-value sensitivity of different tissue types to estimate fibre orientation distributions in each tissue

The following steps are done:
1. Preprocessing
- Denoising
- Unringing
- Motion and distortion correction
- Bias field correction
- Brain mask estimation

2. Estimation of fiber orientation distribution (FOD)
- Response function estimation
- Estimation of FOD
- Intensity Normalization

3. Preparing ATC
- Create a mask with 5 different tissue types (mask for streamline termination)
- Coregistration with DWI
- Create a mask of the gray-matter/white-matter-boundary (mask of streamline seeding)

4. Create streamlines
- Create tractography (10 million streamlines)
- Randomly choose a subset of the 10 million tracks and create a 200k tracks and a 10k tracks
- Filter the tractograms to reduce CSD-based bias in overestimation of longer tracks compared to shorter tracks, and reduce the number of streamlines
- Extract the peaks of a spherical harmonic function in each voxel (could be used for others softawres as TractSeg)

The pipeline is based on B.A.T.M.A.N. tutorial and Andy’s brain book tutorial

Note that this pipeline used the FLIRT registration (FSL) to co-register diffusion and anatomical data as suggested in the B.A.T.M.A.N. tutorial. However it seems to not work perfectly for some data.

Pipeline insight

DWI_whole_brain_tractograpy pipeline combines the following pipelines and processes:

- Dwi preprocessing
- FOD estimation
- Create tissue boundaries
- Create tractography
- Editing tractography
- Filtering tractography
- Spherical harmonic extraction

Mandatory inputs parameters

in_dwi (a string representing an existing file)
Diffusion image to preprocess (valid extensions: [.nii, .nii.gz, .mif]). If a NIfTI is supplied, bvec and bval files will be found automatically.
ex. '/home/username/data/raw_data/DWI.nii'
in_dwi_pe_dir (ap, pa, lr, rl, default value is ap, optional)
Phase encoding direction of the in_diw image:

- ap : Anterior to posterior

- pa: Posterior to anterior

- lr: Left to right

- rl: Right toleft
ex. ap
in_dwi_ro_time (a float, optional)
Total readout time of in_dwi image (in seconds).
ex. 2.0
in_b0_reverse (a string representing an existing file)
b=0 volumes with opposing phase-direction which is to be used exclusively by topup for estimating the inhomogeneity field (valid extensions: [.nii, .nii.gz, .mif]).
ex. '/home/username/data/raw_data/b0_reverse.nii'
in_T1w (a string representing an existing file)
Anatomical image (valid extensions: [.nii, .nii.gz]).
ex. '/home/username/data/raw_data/T1w.nii'

Outputs parameters:

sh_peaks (a string representing a file)
The peaks of a spherical harmonic function in each voxel Each volume corresponds to the x, y & z component of each peak direction vector in turn.
ex. '/home/username/data/derived_data/DWI_denoised_unringed_dwifslpreproc_unbias_wm_odf_norm_peaks.mif'

tracks_10mio (a string representing a file)

The tractography with 10 million streamlines

ex. '/home/username/data/derived_data/DWI_denoised_unringed_dwifslpreproc_unbias_wm_odf_norm_tracto.tck'

tracks_200k (a string representing a file)

The reduce tractography with 200k streamlines

ex. '/home/username/data/derived_data/DWI_denoised_unringed_dwifslpreproc_unbias_wm_odf_norm_tracto_200k.tck'

tracks_10k (a string representing a file)

The reduce tractography with 10k streamlines

ex. '/home/username/data/derived_data/DWI_denoised_unringed_dwifslpreproc_unbias_wm_odf_norm_tracto_10k.tck'

tracks_sift (a string representing a file)
The tractography filtered such that the streamline densities match the FOD lobe integrals
ex. '/home/username/data/derived_data/DWI_denoised_unringed_dwifslpreproc_unbias_wm_odf_norm_tracto_sift.tck'

tracks_sift_10k (a string representing a file)

The reduce sift tractography with 10k streamlines

ex. '/home/username/data/derived_data/DWI_denoised_unringed_dwifslpreproc_unbias_wm_odf_norm_tracto_10k.tck'

T1w_coreg (a string representing a file)
Anatomical image coregistered in DWI space.
ex. '/home/username/data/derived_data/T1w_transformed.mif'

Useful links: MRTrix Tutorial B.A.T.M.A.N.: Basic and Advanced Tractography with MRtrix for All Neurophiles Tutorial Andy’s brain book