Integrated MEG/PET analysis with Brainstorm: a user-friendly toolkit for multimodal datasets
Poster No:
1542
Submission Type:
Abstract Submission
Authors:
Diellor Basha1, Chinmay Chinara2, Anand Joshi2, Raymundo Cassani3, Takfarinas Medani4, Richard Leahy2, Sylvain Baillet5
Institutions:
1McGill University, Montreal, QC, 2University of Southern California, Los Angeles, CA, 3McGill University, Montréal, Canada, 4usc, los angeles, CA, 5Montreal Neurological Institute, Montreal, Quebec
First Author:
Co-Author(s):
Introduction:
Multimodal neuroimaging datasets offer unprecedented opportunities for neuroscience research in complex conditions like Alzheimer's disease (AD). However, despite the surge in the availability of these datasets, which combine structural MRI, PET imaging, and electrophysiological modalities such as MEG and EEG, tools capable of co-analyzing the diverse imaging modalities remain scarce. To address this gap, we have developed a multimodal MEG/PET/MRI pipeline fully integrated into Brainstorm software (Tadel et al., 2011), enabling users to seamlessly co-register and analyze MEG, PET, and MRI data. This module automates key processing steps such as the generation of surface-based protein maps from PET data, allowing integration with MEG source mapping.
Methods:
•Volume processing During importing into Brainstorm, dynamic PET volumes are realigned using SPM's realign and reslice functions. The aligned frames are averaged to obtain a 3D static PET volume that is registered to the subject's MRI using rigid mutual-information-based registration. The registered, aligned PET is then imported into Brainstorm. Within Brainstorm, regions of interest (ROIs), such as the cerebral cortex or cerebellum, are defined using parcellated atlases and segmentations integrated within Brainstorm. ROI-based analysis is used to calculate SUVRs by normalizing PET tracer uptake against a reference ROI (e.g., cerebellum). In parallel, this submodule incorporates wrappers for external software like FreeSurfer and BrainSuite that allow the re-processing of the imported PET volume and feature optional smoothing of PET volumes and partial volume correction (PVC).
•Surface-based analysis The surface-based submodule enables the projection of volumetric PET data onto the cortical surface, facilitating vertex-wise analysis of PET and MEG data in a common framework. The surface projection of PET data is based on the tess2mri interpolation matrix computed by Brainstorm. This matrix establishes weights for volumetric MRI voxels contributing to specific vertices on the cortical surface. Weights range between 0 (no contribution) and 1 (full contribution), enabling a smooth mapping of voxel coordinates to the cortical surface. Intensities from PET voxels, registered to the MRI, are scaled by the interpolation weights, resulting in a weighted projection of PET voxels to their nearest vertex.
•Depth-Weighted Mapping: PET values are mapped at three cortical depths: pial surface, white matter surface and the mid-surface which represents the average surface between the pial and white matter layers. To capture the full cortical thickness and to restrict contaminating contributions from white matter voxels or extracerebral artefacts, the three projections are averaged using a weighted average, with the mid-surface set at 50 % and the pial and white matter surfaces set to 25%, respectively.
•Surface-based analysis The surface-based submodule enables the projection of volumetric PET data onto the cortical surface, facilitating vertex-wise analysis of PET and MEG data in a common framework. The surface projection of PET data is based on the tess2mri interpolation matrix computed by Brainstorm. This matrix establishes weights for volumetric MRI voxels contributing to specific vertices on the cortical surface. Weights range between 0 (no contribution) and 1 (full contribution), enabling a smooth mapping of voxel coordinates to the cortical surface. Intensities from PET voxels, registered to the MRI, are scaled by the interpolation weights, resulting in a weighted projection of PET voxels to their nearest vertex.
•Depth-Weighted Mapping: PET values are mapped at three cortical depths: pial surface, white matter surface and the mid-surface which represents the average surface between the pial and white matter layers. To capture the full cortical thickness and to restrict contaminating contributions from white matter voxels or extracerebral artefacts, the three projections are averaged using a weighted average, with the mid-surface set at 50 % and the pial and white matter surfaces set to 25%, respectively.
Results:
The resulting PET surface maps are integrated with MEG-derived spatio-temporal current density maps, enabling multimodal analysis. Using existing Brainstorm functions for MEG analysis, users can compute spectral power and connectivity features from MEG data, using regional surface ROIs.
Conclusions:
The Brainstorm PET module provides pipelines for integrated, volume- and surface-based analysis of MEG, PET and MRI data.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s)
Modeling and Analysis Methods:
EEG/MEG Modeling and Analysis
Methods Development 1
PET Modeling and Analysis 2
Neuroinformatics and Data Sharing:
Workflows
Keywords:
MEG
Open-Source Software
Positron Emission Tomography (PET)
Workflows
1|2Indicates the priority used for review
