Localization of Realistic Spatial Patches of Complex Source Activity in MEG and EEG

Area of Science:

• Neuroscience
• Biophysics
• Medical Imaging

Background:

• Accurate neural source localization using Magnetoencephalography (MEG) and Electroencephalography (EEG) is critical for neuroscience research and clinical applications.
• Current methods like dipole fitting and distributed source imaging have limitations in accurately capturing sources with complex spatial extents.

Purpose of the Study:

• Introduce PATCH-AP, an enhanced Alternating Projection (AP) method, designed to effectively localize both discrete and spatially extended neural sources.
• Evaluate PATCH-AP's performance against established and recent source localization techniques.

Main Methods:

• PATCH-AP, an enhanced Alternating Projection (AP) method.
• Comparative analysis with distributed source imaging (MNE, sLORETA), dipole fitting (AP), and extended source methods (Convexity-Champagne, FLEX-AP).
• Evaluation using simulations and real MEG data from face perception and auditory tasks.

Main Results:

• PATCH-AP demonstrated superior performance in simulations, achieving lower Earth Mover's Distance (EMD) scores compared to all tested methods.
• Real MEG data analysis showed PATCH-AP aligning well with known brain regions, including the fusiform face area and auditory cortex.
• Consistent outperformance across various source configurations and data types.

Conclusions:

• PATCH-AP offers enhanced accuracy for localizing neural sources with complex spatial characteristics.
• This method holds significant potential for advancing neuroscience research and improving clinical diagnostic capabilities.
• PATCH-AP represents a promising development in the field of brain source imaging.