Localization of Brain Signals by Alternating Projection

Area of Science:

• Neuroscience
• Biophysics
• Biomedical Engineering

Background:

• Magnetoencephalography (MEG) and electroencephalography (EEG) are crucial for modeling brain activity.
• Current dipole models are a popular approach, representing local brain activation.
• Accurate localization of multiple simultaneous sources remains a challenge.

Purpose of the Study:

• To introduce and evaluate a novel method, Alternating Projection (AP), for multiple dipole localization in MEG/EEG.
• To compare the performance of AP against established methods like MUSIC and beamformers.

Main Methods:

• Developed the Alternating Projection (AP) algorithm for multi-dipole source localization.
• Formulated AP to minimize the least-squares (LS) criterion iteratively.
• Localized one dipole at a time while keeping others fixed.

Main Results:

• AP demonstrated high accuracy in simulated, phantom, and human MEG data.
• AP outperformed MUSIC and beamformer scanning methods in localization accuracy.
• AP showed increased robustness to forward model errors and noise.

Conclusions:

• The Alternating Projection (AP) method provides a highly accurate and robust solution for multiple dipole localization in MEG/EEG.
• AP offers significant advantages over conventional scanning methods, especially in challenging conditions like low SNR and correlated sources.