Disco and Dartel complete pipeline

This pipeline allows you to run the whole DISCO + DARTEL registration in one go.

It can also apply the deformation fields to the input data of each subject: sulcal graph, grey-white and pial meshes, and bias-corrected T1-weighted volume.

Finally, it incorporates an Evaluation pipeline to aid in assessing the quality of the resulting registration.

Description

Walkthrough

Please note that the DISCO Study (study_dir) must have been created and populated with a few files prior to running this process. Refer to the DISCO walkthrough for how to create it.

Set template_dir to the DISCO_<Study>/DARTEL_<Template> directory where you want to compute the DISCO+DARTEL coregistration. You should click on the red pile, and fill in the Database, Study, and Template attributes, then hit Enter to make the desired entry appear, before hitting Ok.

The study_dir parameter should be automatically completed, if that is not the case there is an error with template_dir.

Select the subjects that you want to include using the Lgraph attribute. Beware to select the correct version, particularly if you have converted some subjects into disco_analysis.

Note that the 01. Sulcal imprints extraction process will save a file named subjects_hierarchy, which contains the list of subjects used in the Study, and allows autocompletion of this list when you open that Study again.

You may adjust the regularization paramaters of DISCO and/or DARTEL:

For DISCO, go to Disco pipeline / 02. Template building and adjust the value of coef_sigV. This value is expressed in millimeters, and refers to the scale of smoothness that is imposed on the deformation field. Larger values impose more regularity on the result, smaller values allow strong local deformations.

For DARTEL, go to Dartel pipeline / 02. DARTEL Template Creation. There are numerous regularization parameters for DARTEL, which are shortly documented in BrainVISA (hover on the parameter's name to read the tooltip). Please refer the SPM manual for a more complete explanation.

Hit Run! The processing takes a long time (a few hours for few subjects, several days for 50-100 subjects). Note that the slowest part of the pipeline is DARTEL (typically DISCO accounts for about a third of the processing duration, DARTEL for two thirds).

Troubleshooting

If you get an error like Mandatory argument ... has no value, you can try two things:

In the BrainVISA preferences, deactivate databases that are not used by the current Study. Autocompletion of parameters can be upset by extra databases that contain redundant or similar data from the same subjects. Try to reopen the Disco and Dartel complete pipeline from scratch.

Double-check the t1mri_nobias parameter of Disco pipeline: if it has empty values, there is probably an issue with your T1 images not being in NIfTI format.

Special cases

If you use exotic sulci in the registration, i.e. sulci that are not present in the latest nomenclature (sulci_model_2018.trl as of BrainVISA 5.0), you should check and adapt the label_translation parameter. The reason is that DISCO applies a two-step translation of labels:

label_translation translates sulci names using a name-to-name dictionary, which is meant to map fine-grained labels to a given level of granularity in the nomenclature.

sulci_label_translation_siGraph converts the resulting names to integer labels.

Parameters

anatomical_constraints: Choice ( input )
Choose the type of constraints that will drive the registration (Sulci, Bundles, or both). Note that Bundles are an experimental feature, you need to set userLevel to at least Advanced in BrainVISA preferences in order to use this type of constraint.

template_dir: DARTEL Template Dir ( output )
Directory where the DARTEL experiment will be saved. To select a correct value for a new DISCO+DARTEL experiment, you should click on the red pile, and fill in the database, Study, and Template attributes, then hit Enter to make the desired entry appear, before hitting Ok.

study_dir: DISCO Study Dir ( input )
Directory where the DISCO experiment will be saved. Please fill template_dir above, this parameter will be completed automatically.

Lgraph: ListOf( Labelled Cortical folds graph ) ( input )
List of left-side labelled sulcal graphs for all subjects in the DISCO+DARTEL experiment. The other input data (right-side graphs, bias-corrected T1-weighted image, and meshes) are deduced automatically based on this parameter.

label_translation: Label translation ( input )
Dictionary for translating sulcal labels toward a common nomenclature. Keep the default, which corresponds to the nomenclature used for automatic labelling, unless you want to use sulci that are remapped by the default dictionary (can be useful for non-human primate or for experimenting with using small sulci as constraints).

sulci_to_be_studied: Sulci To Be Studied DISCO ( input )
List of sulcal constraints. This list must be created separately using the processes in Preprocessing.

boundingbox_min_mm: ListOf( Float ) ( input )
Minimum coordinates of the common template space, in millimeters relative to the AC point (Anterior Commissure), along the Left, Posterior, Superior axes, respectively (a.k.a. AIMS internal coordinates). Note that the old default was -105.0 -100.0 -110.0. You may need to use this for compatibility with older studies (until BrainVISA 5.0.x).

boundingbox_max_mm: ListOf( Float ) ( input )
Maximum coordinates of the common template space, in millimeters relative to the AC point (Anterior Commissure), along the Left, Posterior, Superior axes, respectively (a.k.a. AIMS internal coordinates).

isotropic_resolution_mm: Float ( input )
Voxel size of the common template space, in millimeters.

Technical information

Toolbox : Disco

User level : 0

Identifier : DISCO_DARTEL_Complete_pipeline

File name : brainvisa/toolboxes/disco/processes/DISCO_DARTEL_Complete_pipeline.py

Supported file formats :
template_dir :
Directory, Directory
study_dir :
Directory, Directory
Lgraph :
Graph and data, Graph and data
label_translation :
Label Translation, Label Translation
sulci_to_be_studied :
Text file, Text file