Statistical evaluation of coherence estimated from optimally beamformed signals

Area of Science:

• Signal Processing
• Biomedical Engineering
• Statistical Analysis

Background:

• Coherence analysis is crucial for understanding signal source interactions.
• Extracting signals through noisy sensor arrays presents significant challenges.
• Existing methods for evaluating coherence estimates are insufficient for noisy, array-processed signals.

Purpose of the Study:

• To develop a method for accurate coherence analysis of signals measured through noisy sensor arrays.
• To address the bias in coherence estimates obtained from optimal beamforming with reference.
• To introduce a novel statistical test for evaluating biased coherence estimates.

Main Methods:

• Optimal beamforming with reference was employed to extract the signal from the sensor array.
• A new statistical test was derived to evaluate the biased coherence estimate.
• The methodology was applied to analyze electroencephalography (EEG) and electromyography (EMG) signals.

Main Results:

• The optimal beamforming with reference approach yields a biased coherence estimate.
• A new statistical test was successfully derived and validated for evaluating this biased estimate.
• The proposed methodology demonstrated advantages over the surface Laplacian for EEG-EMG coherence analysis.

Conclusions:

• The developed statistical test provides a reliable means to evaluate biased coherence estimates in noisy signal scenarios.
• This approach offers significant benefits compared to traditional spatial filters like the surface Laplacian for EEG-EMG coherence analysis.
• The findings enhance the capability for accurate signal source coherence analysis in biomedical applications.