Improving source estimation of retinotopic MEG responses by combining data from multiple subjects

Area of Science:

• Neuroscience
• Biophysics
• Medical Imaging

Background:

• Magnetoencephalography (MEG) measures brain activity via magnetic fields, but source localization is an ill-posed inverse problem.
• Current MEG analysis typically uses individual subject data, potentially limiting spatial resolution.
• Multisubject analysis is proposed to enhance MEG spatial resolution by utilizing intersubject variability.

Purpose of the Study:

• To compare the efficacy of three multisubject analysis methods for improving MEG spatial resolution.
• To evaluate the impact of increasing subject count on the accuracy of MEG source localization.
• To determine which multisubject method offers superior spatial accuracy for retinotopic mapping.

Main Methods:

• Retinotopic mapping data from 20 subjects were analyzed using three multisubject methods: eLORETA with source-space averaging, minimum Wasserstein estimates (MWE), and MWE with source-space averaging.
• Spatial accuracy was quantified by measuring geodesic distances between MEG-derived activation peaks and fMRI-defined retinotopic targets in the primary visual cortex (V1).
• The analysis systematically increased the number of subjects from 1 to 10 to assess the effect of data pooling.

Main Results:

• Increasing the subject count from 1 to 10 reduced median distances between MEG activations and fMRI targets by 6.6-9.4 mm (33-46%), compared to single-subject analyses (approx. 20 mm).
• Multisubject analysis resulted in peak activation locations that better corresponded with established retinotopic maps of V1.
• Averaging eLORETA estimates in source space across individuals demonstrated superior spatial accuracy compared to MWE in this dataset.

Conclusions:

• Pooling data from multiple subjects enhances the spatial accuracy of magnetoencephalography.
• Multisubject analysis, particularly source-space averaged eLORETA, offers improved spatial resolution for functional brain imaging.
• Leveraging intersubject differences through multisubject analysis is a promising strategy for advancing MEG capabilities.