Constrained RLS algorithm for narrow band interference rejection from EEG signal during CES.

Area of Science:

• Biomedical Signal Processing
• Digital Signal Processing
• Neuroscience Instrumentation

Background:

• Narrow-band interference noise is a significant challenge in processing biomedical signals like electroencephalography (EEG).
• Cranial electrical stimulation (CES) can introduce specific narrow-band white Gaussian noise into EEG recordings.
• Existing adaptive filters struggle to efficiently adapt to changing noise characteristics.

Purpose of the Study:

• To develop and evaluate a novel adaptive band-rejection filter for removing double narrow-band noise from EEG signals.
• To improve the filtering performance during cranial electrical stimulation (CES) procedures.
• To introduce a constrained recursive least-squares (CRLS) algorithm that allows true adaptation of filter zeros and poles.

Main Methods:

• Design of multiple adaptive Infinite Impulse Response (IIR) digital band-rejection filters using pole-zero placement on the unit circle.
• Cascading N second-order band-rejection filters to achieve a higher-order filter (2N).
• Utilizing a unique second-order filter structure and convoluting coefficients for filter design.
• Updating filter coefficients via a constrained recursive least-squares (CRLS) algorithm.

Main Results:

• The proposed CRLS multiple adaptive HR band-rejection filter demonstrates true adaptation of both zeros and poles.
• Effective filtering of double narrow-band white Gaussian noise from EEG signals during CES.
• Improved signal quality compared to existing RLS-based adaptive filters.

Conclusions:

• The CRLS adaptive filter provides a robust solution for narrow-band noise removal in EEG signals.
• This method enhances the accuracy of EEG analysis, particularly in the presence of interference from CES.
• The true adaptation of zeros and poles offers superior performance in dynamic noise environments.