[Paper Review] Sparse Bayesian Learning for EEG Source Localization

Sajib Saha, Frank de Hoog+4 more|arXiv (Cornell University)|Jan 19, 2015

Sparse and Compressive Sensing Techniques34 references3 citations

TL;DR

This paper proposes a novel EEG source localization method using Sparse Bayesian Learning (SBL) with an isotropy constraint on current dipoles, grouped within brain regions to enhance sparsity modeling. In a realistic head model, the method achieves >75% accuracy in localizing up to six simultaneous sources, significantly outperforming standard SBL without isotropy, which struggles beyond three sources.

ABSTRACT

Purpose: Localizing the sources of electrical activity from electroencephalographic (EEG) data has gained considerable attention over the last few years. In this paper, we propose an innovative source localization method for EEG, based on Sparse Bayesian Learning (SBL). Methods: To better specify the sparsity profile and to ensure efficient source localization, the proposed approach considers grouping of the electrical current dipoles inside human brain. SBL is used to solve the localization problem in addition with imposed constraint that the electric current dipoles associated with the brain activity are isotropic. Results: Numerical experiments are conducted on a realistic head model that is obtained by segmentation of MRI images of the head and includes four major components, namely the scalp, the skull, the cerebrospinal fluid (CSF) and the brain, with appropriate relative conductivity values. The results demonstrate that the isotropy constraint significantly improves the performance of SBL. In a noiseless environment, the proposed method was 1 found to accurately (with accuracy of >75%) locate up to 6 simultaneously active sources, whereas for SBL without the isotropy constraint, the accuracy of finding just 3 simultaneously active sources was <75%. Conclusions: Compared to the state-of-the-art algorithms, the proposed method is potentially more consistent in specifying the sparsity profile of human brain activity and is able to produce better source localization for EEG.

Motivation & Objective

• To improve the accuracy and consistency of EEG source localization by enhancing sparsity modeling in brain activity.
• To address the challenge of localizing multiple simultaneous neural sources in noisy and ill-posed EEG inverse problems.
• To incorporate physical constraints—specifically isotropy of current dipoles—into the SBL framework for improved source resolution.
• To evaluate the performance of the proposed method against standard SBL in realistic, MRI-based head models.
• To demonstrate the feasibility of localizing up to six active sources with high accuracy using constrained SBL.

Proposed method

• The method employs Sparse Bayesian Learning (SBL) to estimate the locations and strengths of neural current sources from scalp EEG measurements.
• It introduces grouping of current dipoles within brain regions to better reflect the spatial sparsity of neural activity.
• An isotropy constraint is imposed, assuming that the electric current dipoles associated with brain activity are isotropic, reducing model ambiguity.
• The approach uses a realistic head model derived from MRI segmentation, including scalp, skull, CSF, and brain with appropriate conductivity values.
• The inverse problem is solved using a Bayesian framework that promotes sparse solutions while enforcing physical plausibility through the isotropy constraint.
• The algorithm is evaluated numerically using simulated EEG data under controlled noise-free conditions.

Experimental results

Research questions

• RQ1 Can incorporating isotropy constraints into SBL improve the accuracy of EEG source localization in realistic head models?
• RQ2 How does the performance of SBL with dipole grouping and isotropy compare to standard SBL in localizing multiple simultaneous neural sources?
• RQ3 To what extent does the isotropy constraint enhance sparsity modeling and source resolution in EEG inverse problems?
• RQ4 Can the proposed method reliably localize up to six active sources with high accuracy in a noiseless environment?
• RQ5 Does the grouping of dipoles within brain regions improve the robustness and consistency of source localization?

Key findings

• In a noiseless environment, the proposed method accurately localized up to six simultaneously active sources with an accuracy exceeding 75%.
• In contrast, standard SBL without the isotropy constraint achieved less than 75% accuracy even for only three simultaneously active sources.
• The isotropy constraint significantly improved the performance of SBL by reducing solution ambiguity and enhancing sparsity modeling.
• The method demonstrated superior consistency in identifying the sparsity profile of brain activity compared to state-of-the-art algorithms.
• The use of grouped dipoles within brain regions contributed to more realistic and stable source localization results.
• The results indicate that the proposed SBL-based method is a promising approach for high-accuracy, multi-source EEG source localization.