statistics-surface
- Surface-based mass-univariate analysis with SurfStat¶
This command performs statistical analysis (e.g. group comparison, correlation) on surface-based features using the GLM.
Warning
Prior to release 0.7.3
of Clinica, this pipeline was relying on the Matlab toolbox SurfStat
designed for statistical analyses of univariate and multivariate surface and volumetric data using the GLM [Worsley et al., 2009].
SurfStat, as a Matlab toolbox, necessitates a valid Matlab license to be executed, which could be a heavy dependency for users.
Moreover, the maintenance of SurfStat ceased a few years ago which posed some issues, especially with newer versions of Matlab.
For these reasons, the pipeline was completely rewritten to rely on Brainstat
which is a pure Python implementation of SurfStat, offering a very similar API.
The different images produced by the pipeline are generated using Nilearn, also a pure Python library.
Although the dependencies have been reduced, please be aware that this pipeline has not been extensively tested with the new implementation.
Do not hesitate to open a new issue on GitHub to report bugs you might encounter.
Surface-based measurements are analyzed on the FsAverage surface template (from FreeSurfer).
Currently, this pipeline can handle:
- cortical thickness measurements from T1 images
t1-freesurfer
pipeline - map of activity from PET data using
pet-surface
pipeline
Prerequisites¶
You need to process your data with the t1-freesurfer
pipeline for measurements of cortical thickness measurements from T1 images or pet-surface
pipeline for measurements of activity map from PET.
Do not hesitate to have a look at the paragraph Specifying what surface data to use if you want to use your own surface feature.
Dependencies¶
If you only installed the core of Clinica, this pipeline needs the installation of FreeSurfer on your computer.
Running the pipeline¶
The pipeline can be run with the following command line:
clinica run statistics-surface [OPTIONS] CAPS_DIRECTORY GROUP_LABEL {t1-freesurfer|pet-surface|custom-pipeline}
{group_comparison|correlation} SUBJECT_VISITS_WITH_COVARIATES_TSV CONTRAST
where:
CAPS_DIRECTORY
is the folder containing the results of thet1-freesurfer
orpet-surface
pipeline and the output of the present command, both in a CAPS hierarchy.GROUP_LABEL
is a string defining the group label for the current analysis, which helps you keep track of different analyses.- The third positional argument defines the type of surface-based feature. It can be:
t1-freesurfer
for cortical thicknesspet-surface
for projected PET datacustom-pipeline
for you own data in CAPS directory (see below for details)
- The fourth positional argument is a string defining the type of analysis of your model. It can be either
group_comparison
orcorrelation
. SUBJECT_VISITS_WITH_COVARIATES_TSV
is a TSV file containing a list of subjects with their sessions and all the covariates and factors in your model (the content of the file is explained in the Example subsection).CONTRAST
is a string defining the contrast matrix or the variable of interest for the GLM. For examplegroup
,sex
, orage
.
Pipeline options:
-c
, or--covariates
: Covariates must be provided one at the time:-c covariate_1 -c covariate_2
, or equivalently--covariates covariate_1 --covariates covariate_2
. Each covariate must match the name of the TSV file. By default, no covariate is considered.--full_width_at_half_maximum
: FWHM for the surface smoothing. The default value is20
.
Pipeline options if you use inputs from the pet-surface
pipeline:
--acq_label
: Name of the label given to the PET acquisition, specifying the tracer used (trc-<acq_label>
).--suvr_reference_region
: Reference region used to perform intensity normalization (i.e. dividing each voxel of the image by the average uptake in this region) resulting in a standardized uptake value ratio (SUVR) map. It can be eithercerebellumPons
(used for amyloid tracers), orpons
(used for FDG).
Tip
Check the Example subsection for further clarification.
Outputs¶
Group comparison analysis¶
Results are stored in the following folder of the CAPS hierarchy: groups/<group_id>/statistics/surfstat_group_comparison/
.
The main outputs for the group comparison are:
<group_id>_<group_1>-lt-<group_2>_measure-<label>_fwhm-<label>_correctedPValue.jpg
: contains both the cluster level and the vertex level corrected p-value maps, based on the random field theory.<group_id>_<group_1>-lt-<group_2>_measure-<label>_fwhm-<label>_FDR.jpg
: contains corrected p-value maps, based on the false discovery rate (FDR).
The <group_1>-lt-<group_2>
means that the tested hypothesis is: "the measurement of <group_1>
is lower than (lt
) the measurement of <group_2>
".
The pipeline includes both contrasts so *<group_2>-lt-<group_1>*
files are also saved.
The FWHM value corresponds to the size in mm of the kernel used to smooth the surface and can be 5, 10, 15, 20.
Analysis with cortical thickness (respectively PET data) will be saved under the _measure-ct
keyword (respectively the _measure-<acq_label>
keyword).
Correlations analysis¶
Results are stored in the following folder of the CAPS hierarchy: groups/<group_id>/statistics/surfstat_correlation/
.
The main outputs for the correlation are:
<group_id>_correlation-<label>_contrast-<label>_measure-<label>_fwhm-<label>_correctedPValue.jpg
: contains p-value maps corrected at both the cluster and vertex levels, based on the random field theory.<group_id>_correlation-<label>_contrast-<label>_measure-<label>_fwhm-<label>_FDR.jpg
: contains corrected p-value maps, based on the false discovery rate (FDR).<group_id>_correlation-<label>_contrast-<label>_measure-<label>_fwhm-<label>_T-statistics.jpg
: contains the maps of T statistics.<group_id>_correlation-<label>_contrast-<label>_measure-<label>_fwhm-<label>_Uncorrected p-value.jpg
: contains the maps of uncorrected p-values.
The correlation-<label>
describes the factor of the model, which can be for example age
.
The contrast-<label>
is the sign of your factor which can be negative
or positive
.
Analysis with cortical thickness (respectively PET data) will be saved under the _measure-ct
keyword (respectively the _measure-<acq_label>
keyword).
Note
The full list of output files can be found in the ClinicA Processed Structure (CAPS) specifications.
Example¶
Comparison analysis¶
Let's assume that you want to perform a group comparison of cortical thickness between patients with Alzheimer’s disease (group_1
will be called AD
) and healthy subjects (group_2
will be called HC
).
ADvsHC
will define the group_label
.
The TSV file containing the participants and covariates will look like this:
participant_id session_id sex group age
sub-CLNC0001 ses-M000 Female CN 71.1
sub-CLNC0002 ses-M000 Male CN 81.3
sub-CLNC0003 ses-M000 Male CN 75.4
sub-CLNC0004 ses-M000 Female CN 73.9
sub-CLNC0005 ses-M000 Female AD 64.1
sub-CLNC0006 ses-M000 Male AD 80.1
sub-CLNC0007 ses-M000 Male AD 78.3
sub-CLNC0008 ses-M000 Female AD 73.2
We call this file ADvsHC_participants.tsv
.
For this group comparison, we will use age
and sex
as covariates.
As a result, the command line will be:
clinica run statistics-surface caps_directory ADvsHC t1-freesurfer group_comparison ADvsHC_participants.tsv group -c age -c sex
The results of the group comparison between AD and HC are given by the group-ADvsHC_AD-lt-HC_measure-ct_fwhm-20_correctedPValue.jpg
file.
Correlation analysis¶
Let's now assume that you are interested in knowing whether cortical thickness is correlated with age using the same population as above, namely ADvsHC_participants.tsv
.
The contrast will become age
and we will choose correlation
instead of group_comparison
. The command line is:
clinica run statistics-surface caps_directory ADvsHC t1-freesurfer correlation ADvsHC_participants.tsv age -c group -c sex
Describing this pipeline in your paper¶
Example of paragraph (group comparison):
These results have been obtained using the statistics-surface
command of Clinica [Routier et al., 2021].
More precisely, a point-wise, vertex-to-vertex model based on the Brainstat Python library (https://brainstat.readthedocs.io/en/master/) was used to conduct a group comparison of whole brain cortical thickness.
The data were smoothed using a Gaussian kernel with a full width at half maximum (FWHM) set to <FWHM>
mm.
The general linear model was used to control for the effect of <covariate_1>
, ... and <covariate_N>
.
Statistics were corrected for multiple comparisons using the random field theory for non-isotropic images [Worsley et al., 1999].
A statistical threshold of P < <ClusterThreshold>
was first applied (height threshold).
An extent threshold of P < 0.05 corrected for multiple comparisons was then applied at the cluster level.
Tip
Easily access the papers cited on this page on Zotero.
Support¶
- You can use the Clinica Google Group to ask for help!
- Report an issue on GitHub.
(Advanced) Specifying what surface data to use¶
The optional flag --custom_file CUSTOM_FILE
allows you to specify yourself what file should be taken in the CAPS/subjects
directory.
CUSTOM_FILE
is a string describing the folder hierarchy to find the file.
For instance, let's say we want to manually indicate to use the cortical thickness.
Here is the generic link to the surface data files:
CAPS/subjects/sub-*/ses-M*/t1/freesurfer_cross_sectional/sub-*_ses-M*/surf/*h.thickness.fwhm*.fsaverage.mgh
Example: CAPS/subjects/sub-ADNI011S4075/ses-M000/t1/freesurfer_cross_sectional/sub-ADNI011S4075_ses-M000/surf/lh.thickness.fwhm15.fsaverage.mgh
Note that the file must be in the CAPS/subjects
directory.
So my CUSTOM_STRING
must only describe the path starting after the subjects
folder. We need to replace the *
by the correct keywords, in order for the pipeline to catch the correct filenames:
@subject
is the subject@session
is the session@hemi
is the hemisphere@fwhm
is the full width at half maximum
All those variables are already known, you just need to indicate where they are in the filename! As a result, we will get for CUSTOM_FILE
of cortical thickness:
@subject/@session/t1/freesurfer_cross_sectional/@subject_@session/surf/@hemi.thickness.fwhm@fwhm.fsaverage.mgh
You will finally need to define the name your surface feature --feature_label FEATURE_LABEL
. It will appear in the _measure-<FEATURE_LABEL>
of the output files once the pipeline has run.
Appendix¶
- For more information about Brainstat, please check here.
- For more information about the GLM, please check here.
- The cortical thickness map is obtained from the FreeSurfer segmentation. More precisely, it corresponds to the subject’s map normalized onto FSAverage and smoothed using a Gaussian kernel FWHM of
<fwhm>
mm (thesurf/?h.thickness.fwhm<fwhm>.fsaverage.mgh
files).
Contact us !¶
- Check for past answers on Clinica Google Group
- Start a discussion on GitHub
- Report an issue on Github