Medial Tractography Analysis (MeTA)

MeTA is a workflow implemented to minimize microstructural heterogeneity in diffusion MRI (dMRI) metrics by extracting and parcellating the core volume along the bundle length in the voxel-space directly while effectively preserving bundle shape and efficiently capturing the regional variation within and along white matter (WM) bundles.

Contact: Iyad Ba Gari iyad.bagari@usc.edu

If you use MeTA code, please cite the following publication:

• Ba Gari, I., et al.: Heritability and Genetic Correlations Along the Corticospinal Tract. International Workshop on Computational Diffusion MRI. Cham: Springer Nature Morocco, 2024
• Ba Gari, I., et al.: Medial tractography analysis (MeTA) for white matter population analyses across datasets. In: 2023 11th International IEEE/EMBS Conference on Neural Engineering (NER). pp. 1–5 (Apr 2023)

Installation

There are two options to use the package: via Conda or Docker/Singularity.

Conda Installation

Create an environment with Python version >=3.9 and <3.12. For example:

conda config --add channels bioconda
conda create -n meta python==3.13
conda activate meta
conda install bioconda::meta-neuro=2.0.1

Singularity Installation

To pull the singularity image using apptainer:

apptainer pull meta_2_0_1.sif docker://quay.io/biocontainers/meta-neuro:2.0.1--py313h47f2c4e_0

apptainer run meta_2_0_1.sif meta --help

NOTE: Use meta --help to see the package options.

How to use the package:

Generate Medial Surface for WM Bundle:

Medial surface is extract based on Continuous medial representation (CMREP) method Yushkevich, 2009.

• Convert streamlines in trk/tck/tt.gz formats to a binary image.

density_map --tractogram CST.trk --reference dti_FA.nii.gz --output CST.nii.gz

• Generate a 3D Medial Surface for WM Bundle using the CMREP Method:

vtklevelset CST.nii.gz CST.vtk 0.1
cmrep_vskel -c 3 -p 1.5 -g CST.vtk CST_skeleton.vtk

Run Medial Tractography Analysis (MeTA):

MeTA will extract the core volume of WM bundle and parcellate it into segments along the bundle length.

meta --subject 1234 --bundle CST --medial_surface CST_skeleton.vtk --volume CST.vtk --sbundle CST.trk --mbundle CST_model.trk --transform subject_ANTs0GenericAffine.mat --mask CST.nii.gz --num_segments 15 --output CST

Extract Voxel-based Bundle Profile:

Compute volumetric profile based on binary masks and microstructure maps e.g., FA, MD, RD, AD, etc. Output two files: 1) *_segments_average.csv file with the average profile along the bundle length, and 2) *_segments_voxelwise.h5: the profile for each voxel in the bundle.

volumetric_profile --subject 1234 --bundle CST --mask CST_local_all.nii.gz --map FA.nii.gz --output /output_folder

Extract Streamline-based Profile:

Compute streamline profile based on tractography and microstructure maps e.g., FA, MD, RD, AD, etc. output two files: 1) *_streamlines_average.csv file with the average profile along the bundle length, and 2) *_streamlines_pointwise.h5: the profile for each point of streamline.

streamlines_profile --subject 1234 --bundle CST --tractogram CST.trk --mask CST_local_all.nii.gz --map FA.nii.gz --output /output_folder

Extract Bundle Shape Features:

Bundle shape features implemented based on Yeh et al., 2020. The following features are extracted:

1. Total number of streamlines
2. Average streamlines length
3. Span
4. Curl
5. Volume
6. Surface area
7. Diameter
8. Elongation
9. Irregularity

shape_metrics --subject 1234 --bundle CST --mask CST.nii.gz --tractogram CST.trk --output CST_streamlines_metrics.csv