Time-frequency mixed-norm estimates: sparse M/EEG imaging with non-stationary source activations.

Area of Science:

• Neuroscience
• Biophysics
• Signal Processing

Background:

• Magnetoencephalography (MEG) and electroencephalography (EEG) provide high temporal resolution functional brain imaging.
• Current inverse problem solutions often assume signal stationarity, limiting accuracy for transient brain activity.

Purpose of the Study:

• To introduce time-frequency mixed-norm estimates (TF-MxNE) for improved functional brain imaging.
• To address the limitations of stationary signal assumptions in existing inverse solvers.

Main Methods:

• Developed TF-MxNE utilizing time-frequency analysis for regularizing ill-posed inverse problems.
• Employed structured sparse priors in the time-frequency domain to capture signal dynamics.
• Utilized convex optimization and proximal operators for an efficient estimation algorithm.

Main Results:

• TF-MxNE accurately captures the non-stationary and transient nature of brain signals.
• Demonstrated improved accuracy compared to existing inverse solvers through simulations.
• Validated the method on publicly available MEG datasets.

Conclusions:

• TF-MxNE offers a significant advancement in functional brain imaging by accounting for temporal signal dynamics.
• The method provides more accurate source localization for complex brain activity.
• This approach enhances the utility of MEG and EEG for neuroscience research.