A multiple coherence-based detector for evoked responses in the EEG during sensory stimulation

Area of Science:

• Neuroscience
• Signal Processing
• Statistical Analysis

Background:

• Evoked responses in electroencephalogram (EEG) signals are often obscured by noise.
• Accurate detection of these responses is crucial for understanding brain activity during sensory stimulation.

Purpose of the Study:

• To derive the sampling distribution of the multiple coherence estimate for periodic signals in noise.
• To establish a statistical framework for detecting evoked responses using this estimate.

Main Methods:

• Derivation of the sampling distribution for multiple coherence estimates in the zero-coherence case.
• Analytical derivation of bias and variance for the estimate.
• Application to EEG data analysis for evoked response detection.

Main Results:

• The probability density function of the estimate varies with the number of data segments for a fixed number of signals.
• Analytical expressions for bias and variance were obtained.
• The derived statistical apparatus, including critical values, enables robust detection.

Conclusions:

• The developed multiple coherence estimation technique provides a statistically sound method for detecting evoked responses.
• The approach is validated through its application to electroencephalogram (EEG) data during sensory stimulation.