Related Experiment Video
Updated: Jul 10, 2026
08:45
Mapping Cortical Dynamics Using Simultaneous MEG/EEG and Anatomically-constrained Minimum-norm Estimates: an Auditory Attention Example
Published on: October 24, 2012
Systematic regularization of linear inverse solutions of the EEG source localization problem
Christophe Phillips1, Michael D Rugg, Karl J Fristont
1Institute of Cognitive Neuroscience University College London, London, United Kingdom.
Neuroimage
|December 17, 2002
Summary
This study introduces a new method for EEG source localization using anatomical and physiological constraints. The approach adaptively estimates regularization hyperparameters, improving accuracy with varying noise and source geometry.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Distributed linear solutions are standard for EEG source localization.
- Existing methods often lack robust regularization, especially with noisy data.
Purpose of the Study:
- To develop a novel EEG source localization method incorporating anatomical and physiological constraints.
- To systematically estimate regularization hyperparameters using restricted maximum likelihood (ReML).
Main Methods:
- Weighted minimum norm method with anatomical/physiological priors.
- Restricted maximum likelihood (ReML) for hyperparameter estimation.
- Iterative expectation-maximization procedure for joint source distribution and hyperparameter estimation.
Main Results:
- Solutions in informed basis function spaces show high validity compared to conventional methods.
- Regularization hyperparameters significantly vary with source geometry and noise levels.
- Adaptive ReML hyperparameter estimation is crucial for accurate EEG source localization.
Conclusions:
- The proposed method offers a valid approach to EEG source localization by integrating multimodal information.
- Adaptive estimation of regularization hyperparameters is essential for robust performance across different conditions.
