Source imaging of deep-brain activity using the regional spatiotemporal Kalman filter

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Human brain networks involve complex interactions between cortical and subcortical regions.
• Surface electroencephalography (EEG) has limited spatial resolution, hindering the detection of subcortical brain activity.
• Existing EEG source imaging methods often fail to reconstruct signals from deep brain structures due to signal weakening and artifact contamination.

Purpose of the Study:

• To introduce a novel EEG source imaging method, the regional spatiotemporal Kalman filter (RSTKF).
• To demonstrate RSTKF's capability in detecting and accurately localizing deep brain activity.
• To compare RSTKF's performance against established methods like LORETA and standard STKF.

Main Methods:

• The regional spatiotemporal Kalman filter (RSTKF) is a state-space model that accounts for spatially heterogeneous dynamical noise variances.
• RSTKF was validated using simulated EEG data with sources in the frontal lobe, putamen, and thalamus.
• The method was applied to non-averaged interictal epileptic spikes from a patient with focal epilepsy.

Main Results:

• RSTKF successfully and accurately localized deep brain sources in both simulated and real patient data.
• The method demonstrated superior spatial resolution compared to LORETA and standard STKF.
• RSTKF proved effective even with non-averaged epileptic spikes, overcoming signal-to-noise ratio limitations.

Conclusions:

• RSTKF offers accurate, focal, and consistent source localization, especially in deeper brain areas.
• This method holds potential for localizing the epileptogenic zone in deep brain structures for epilepsy surgery.
• RSTKF may be particularly valuable for EEG recordings where averaging is not feasible due to insufficient spike data.