ENIGMA TBSS protocol

TBSS following the ENIGMA protocol (https://enigma.ini.usc.edu/protocols/dti-protocols/).

INPUTS

Subject FA image
Subject MD image

OUTPUTS

fa.nii.gz                          Masked input images  (subject native space)
md.nii.gz

mask.nii.gz                        Brain mask (subject native space)

fa_reg0GenericAffine.mat           ANTS transform from native to template space
fa_reg1Warp.nii.gz

tbss.pdf                           Registration QC PDF

fa_regWarped.nii.gz                Subject FA, MD images (template space)
md_regWarped.nii.gz

fa_regWarped_skeletonised.nii.gz   Skeletonized subject FA, MD images
md_regWarped_skeletonised.nii.gz        (template space)

extracted_roi_means.csv            FA, MD values in JHU ROIs

Preprocessing

Diffusion images are assumed already preprocessed to produce FA and MD images. A typical preprocessing pipeline would be PreQual.

The provided MD image is thresholded at the md_threshold value to produce a tight brain mask, which is then applied to the FA image and also used in the registration.

ENIGMA-DTI Skeletonization

The outline of the ENIGMA protocol is followed (enigma/ENIGMA_TBSS_protocol_USC.pdf). The numbered steps of the protocol are handled as follows:

(1, template files) The default ENIGMA template FA image, tract skeleton, and skeleton distance image are used. These are found in the enigma directory in this repository, but the original source was http://enigma.ini.usc.edu/wp-content/uploads/2013/02/enigmaDTI.zip (downloaded Dec 2024).

(2, file handling only)

(3, FA image erosion) This step is skipped as the registration algorithm does not need it.

(4, registration to template) The single subject FA image is registered to the FA template using ANTS, specifically antsRegistrationSyN.sh. Source code, Documentation. Relevant references for this script include:

• http://www.ncbi.nlm.nih.gov/pubmed/20851191
• http://www.frontiersin.org/Journal/10.3389/fninf.2013.00039/abstract

The registration is also applied to the subject MD image. For QC, a PDF is produced to show the quality of registration between subject and template.

(5, template mask creation) Skipped. The default ENIGMA mask is used.

(6, distance map creation) Skipped. The default ENIGMA skeleton and distance image are used.

(7, parallelization) Not used, as only a single subject is processed.

(8, Skeletonize subject images) FSL's tbss_skeleton function is used to project the registered subject FA and MD images onto the template skeleton.

ROI signal extraction

The ENIGMA protocol is described in enigma/ENIGMA_ROI_protocol_USC.pdf, and the accompanying files are found in the enigma/ROIextraction_info directory of this repository (downloaded Dec 2024). But, the described code and processing steps are not used. The underlying procedure however is followed - Subject FA within the skeleton is averaged within each ROI of the enigma/ROIextraction_info/JHU-WhiteMatter-labels-1mm.nii.gz image, ignoring zero-valued (non-skeleton) voxels. The ROIs are listed in the file enigma/ROIextraction_info/ENIGMA_look_up_table.txt.