Stable EEG source estimation for standardized Kalman filter using rate-of-change tracking

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Accurate localization of brain activity is crucial for presurgical evaluation and treatment planning in neurological diseases like epilepsy.
• Brain activity's dynamic nature necessitates advanced spatiotemporal modeling for precise time-varying behavior capture.

Purpose of the Study:

• To develop an improved parameter tuning and model for the Standardized Kalman filter to enhance the stability and accuracy of brain activity localization.
• To introduce a backward-differentiation-based measurement model for the rate of change in brain activity distribution.

Main Methods:

• A novel parameter tuning and model utilizing the rate-of-change of brain activity distribution was developed.
• Simulated and real non-invasive electroencephalography (EEG) data with realistic head models were used for experiments.
• The proposed method was compared against the original Standardized Kalman filter and Standardized Low-resolution Brain Electromagnetic Tomography (sLORETA).

Main Results:

• The proposed parametrization achieved high localization accuracy, identifying 7 out of 8 expected originators of short-latency somatosensory evoked potentials.
• The rate-of-change model demonstrated superior tracking stability compared to filtering without it, especially against parameter changes.
• The method proved more robust against poorly set model parameters.

Conclusions:

• The enhanced Standardized Kalman filter provides accurate and stable estimations for brain activity localization and dynamical properties.
• The method's success in sub-thalamic localization highlights its potential for guiding stereo-electroencephalography (SEEG) sensor placement and deep-brain stimulation electrode implantation.