t1-linear
- Affine registration of T1w images to the MNI standard space¶
This pipeline performs a set of steps in order to affinely align T1-weighted MR images to the MNI space using the ANTs software package [Avants et al., 2014].
These steps include:
- bias field correction using N4ITK [Tustison et al., 2010]
- affine registration to the MNI152NLin2009cSym template [Fonov et al., 2011, 2009] in MNI space with the SyN algorithm [Avants et al., 2008]
- cropping of the registered images to remove the background
This pipeline was designed as a prerequisite for the `extract and deep learning classification algorithms presented in [Wen et al., 2020].
Dependencies¶
If you only installed the core of Clinica, this pipeline needs the installation of ANTs on your computer.
Tip
Since clinica 0.9.0
you have the option to rely on ANTsPy
instead of ANTs to run this pipeline, which means that the installation of ANTs is not
required in this case. The ANTsPy package is installed with other Python dependencies of Clinica.
To use this options, you simply need to add the --use-antspy
option flag to the command line (see below).
Note however that this is a new and not extensively tested option such that bugs or unexpected
results are possible. Please contact the Clinica developer team if you encounter issues.
Running the pipeline¶
The pipeline can be run with the following command line:
clinica run t1-linear [OPTIONS] BIDS_DIRECTORY CAPS_DIRECTORY
where:
BIDS_DIRECTORY
is the input folder containing the dataset in a BIDS hierarchy.CAPS_DIRECTORY
is the output folder containing the results in a CAPS hierarchy.
On default, cropped images (matrix size 169×208×179, 1 mm isotropic voxels) are generated to reduce the computing power required when training deep learning models.
Use the option --uncropped_image
if you do not want to crop the image.
It is also possible to obtain a deterministic result by setting the value of the random seed used by ANTs with the option --random_seed
. Default will lead to a non-deterministic result.
This option requires ANTs version 2.3.0
onwards. It is also compatible with ANTsPy.
It is possible to specify the name of the CAPS dataset that will be created to store the outputs of the pipeline. This works if this CAPS dataset does not exist yet, otherwise the existing name will be kept.
This can be achieved with the --caps-name
option. The provided name will appear in the dataset_description.json
file, at the root of the CAPS folder (see CAPS Specifications for more details).
Finally, it is possible to use ANTsPy instead of ANTs by passing the --use-antspy
flag.
Note
The arguments common to all Clinica pipelines are described in Interacting with clinica.
Tip
Do not hesitate to type clinica run t1-linear --help
to see the full list of parameters.
Outputs¶
Results are stored in the following folder of the CAPS hierarchy: subjects/<participant_id>/<session_id>/t1_linear
with the following outputs:
<source_file>_space-MNI152NLin2009cSym_res-1x1x1_T1w.nii.gz
: T1w image affinely registered to theMNI152NLin2009cSym
template.- (optional)
<source_file>_space-MNI152NLin2009cSym_desc-Crop_res-1x1x1_T1w.nii.gz
: T1w image registered to theMNI152NLin2009cSym
template and cropped. <source_file>_space-MNI152NLin2009cSym_res-1x1x1_affine.mat
: affine transformation estimated with ANTs.
Going further¶
You can now use the ClinicaDL framework presented in [Wen et al., 2020] for classification or registration quality check based on deep learning methods.
Describing this pipeline in your paper¶
Example of paragraph
These results have been obtained using the t1-linear
pipeline of Clinica
[Routier et al., 2021;
Wen et al., 2020].
More precisely, bias field correction was applied using the N4ITK method
[Tustison et al., 2010].
Next, an affine registration was performed using the SyN algorithm
[Avants et al., 2008]
from ANTs [Avants et al., 2014]
to align each image to the MNI space with the ICBM 2009c nonlinear symmetric template
[Fonov et al., 2011,
2009].
(Optional) The registered images were further cropped to remove the background
resulting in images of size 169×208×179, with 1 mm isotropic voxels.
Tip
Easily access the papers cited on this page on Zotero.
Contact us !¶
- Check for past answers on Clinica Google Group
- Start a discussion on GitHub
- Report an issue on Github