Source Imaging Method Based on Spatial Smoothing and Edge Sparsity (SISSES) and Its Application to OPM-MEG

Area of Science:

• Neuroscience
• Biophysics
• Signal Processing

Background:

• Magnetoencephalography (MEG) source estimation is crucial for understanding brain function but is an ill-posed problem.
• Traditional methods have limitations in accurately estimating source extent and amplitude.
• Existing structured sparsity algorithms show promise but suffer from amplitude bias.

Purpose of the Study:

• To introduce a novel method, SISSES (Source Imaging based on Spatial Smoothing and Edge Sparsity), for improved MEG source estimation.
• To enhance the spatiotemporal resolution of reconstructed neural sources.
• To address the amplitude bias issue present in previous methods.

Main Methods:

• Developed SISSES, modeling temporal dynamics with basis functions and spatial characteristics with a first-order Markov Random Field (MRF).
• Applied sparse constraints to MRF model residuals in original and variation domains.
• Validated SISSES using numerical simulations and real MEG data from auditory and median nerve stimulation.

Main Results:

• SISSES outperformed benchmark methods in estimating the time course, location, and extent of simulated patch sources.
• Real MEG data analysis demonstrated SISSES's ability to accurately identify brain response source regions over time.
• The method showed feasibility for practical applications in MEG analysis.

Conclusions:

• SISSES offers superior spatiotemporal resolution for MEG source imaging compared to existing techniques.
• The novel approach effectively mitigates amplitude bias, leading to more accurate source reconstruction.
• SISSES is a promising tool for advancing the study of brain activity and clinical applications like preoperative functional localization.