Multiple Dipole Sources Localization from the Scalp EEG Using a High-resolution Subspace Approach.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Subspace source localization methods like MUSIC and RAP-MUSIC face challenges with closely-spaced neural sources.
• Accurate localization and resolution are crucial for understanding brain activity.

Purpose of the Study:

• To introduce and evaluate a novel algorithm, FINE (Fictitious Iterative Network Estimation), for enhanced spatial resolution and localization accuracy.
• To compare FINE's performance against established algorithms (MUSIC, RAP-MUSIC) in simulated and real-world neural data.

Main Methods:

• FINE was developed within the subspace source localization framework.
• Performance was evaluated using computer simulations in a realistic head model (Boundary Element Method - BEM).
• FINE was applied to analyze motor potentials from human finger movements.

Main Results:

• FINE successfully distinguished superficial sources as close as 8.5 mm and deep sources as close as 16.3 mm.
• FINE demonstrated superior accuracy in source orientation estimation for closely-spaced sources compared to MUSIC and RAP-MUSIC.
• FINE revealed detailed neural activity distribution in premotor and supplementary motor areas (SMA) during finger movements, surpassing MUSIC's resolution.

Conclusions:

• FINE offers excellent spatial resolution for imaging neural sources, particularly for closely-spaced and complex activity patterns.
• The algorithm provides more accurate source localization and orientation estimation than conventional subspace methods.
• FINE shows significant potential for advancing neuroimaging and understanding brain function.