pet-volume – Pre-processing of PET images for volume analyses¶
This pipeline performs several pre-processing steps on PET data in voxel space, which include:
- intra-subject registration of the PET image into the space of the subject’s T1-weighted MRI image using SPM;
- (optional) partial volume correction (PVC) using the PETPVC toolbox;
- inter-subject spatial normalization of the PET image into MNI space based on the DARTEL deformation model of SPM;
- intensity normalization using the average PET uptake in reference regions resulting in a standardized uptake value ratio (SUVR) map;
- parcellation into anatomical regions based on an atlas and computation of average values within each region. The list of available atlases can be found here.
If you only installed the core of Clinica, this pipeline needs the installation of SPM and CAT12. You can find how to install these software packages on the third-party page.
If you want to apply partial volume correction (PVC) on your PET data, you will need to install PETPVC, which depends on ITK. More information on the third-party page.
Running the pipeline¶
The pipeline can be run with the following command line:
clinica run pet-volume bids_directory caps_directory group_id
bids_directoryis the input folder containing the dataset in a BIDS hierarchy.
caps_directoryacts both as an input folder (where the results of the
t1-volume-*pipeline are stored) and as the output folder containing the results in a CAPS hierarchy.
group_idis the id of the group that is associated to the DARTEL template that you had created when running the
-pet: type of PET image to process. Possible values are fdg and av45. Default value is fdg.
-smooth: a list of integers specifying the different isomorphic full width at half maximum (fwhm) in millimeters to smooth the image. Default value is: 0, 8 (both without smoothing and with an isomorphic smoothing of 8 mm)
-pvc_fwhm: TSV file containing the fwhm_x, fwhm_y and fwhm_z of the PSF for each PET image. More explanation below.
Partial volume correction
To correct for partial volume effects, the pipeline uses the region-based voxel-wise (RBV) correction implemented in the PETPVC toolbox. You need to specify in a TSV file the full width at half maximum (FWHM), in millimeters, of the point spread function (PSF) associated with your data, in the x, y and z directions. For instance, if the FWHM of the PSF associated with your first image is 8 mm along the x axis, 9 mm along the y axis, and 10 mm along z axis, the first row of your TSV file will look like this:
participant_id session_id fwhm_x fwhm_y fwhm_z sub-CLNC0001 ses-M00 8 9 10 sub-CLNC0002 ses-M00 7 6 5 sub-CLNC0003 ses-M00 6 6 6
The arguments common to all Clinica pipelines are described in Interacting with clinica.
Do not hesitate to type
pipeline --help to see the full list of parameters.
Results are stored in the following folder of the CAPS hierarchy:
The main output files are:
<source_file>_space-T1w[_pvc-rbv]_pet.nii.gz: PET image registered into the T1w native space.
<source_file>_space-Ixi549Space[_pvc-rbv]_suvr-<label>_mask-brain[_fwhm-<X>mm]_pet.nii.gz: standard uptake value ratio (SUVR) PET image in MNI space, masked to keep only the brain, and optionally smoothed.
atlas_statistics/<source_file>_space-<space>[_pvc-rbv]_suvr-<label>_statistics.tsv: TSV files summarizing the regional statistics on the labelled atlas <space>.
The [pvc-rbv] label indicates whether the PET image has undergone partial value correction (region-based voxel-wise method) or not.
The full list of output files from the pet-volume pipeline can be found in the The ClinicA Processed Structure (CAPS) Specification.
Describing this pipeline in your paper¶
Example of paragraph:
Theses results have been obtained using the
pet-volume pipeline of Clinica. This pipeline first performs intra-subject registration of the PET image into the space of the subject’s T1-weighted MRI image using SPM. [The PET image is corrected for partial volume effects using the PETPVC toolbox.] The PET image is then spatially normalized into MNI space using the DARTEL deformation model of SPM, and intensity normalized using the average PET uptake in a reference region ([pons | pons + cerebellum]). Finally, the PET image is parcellated into anatomical regions based on an atlas, and average values are computed within each region.