Comparison of time-variant coherence algorithms in single-trial: a dynamic analysis

Area of Science:

• Neuroscience
• Signal Processing
• Computational Biology

Background:

• Brain region communication is crucial for cognitive functions like learning and memory.
• Electroencephalogram (EEG) synchronization, quantified by time-variant coherence, offers insights into neural communication.
• Analyzing single-trial EEG data presents challenges for accurately measuring dynamic brain activity.

Purpose of the Study:

• To compare the performance of three distinct time-variant coherence algorithms for single-trial EEG analysis.
• To evaluate algorithms based on short-time Fourier transform, adaptive discrete Fourier transform, and recursive smoothed pseudo Wigner-distribution.
• To identify differences in estimation performance and discuss suitability for future neuroscience applications.

Main Methods:

• Implementation of three time-variant coherence algorithms: short-time Fourier transform (STFT), adaptive discrete Fourier transform (ADFT), and recursive smoothed pseudo Wigner-distribution (SPWD).
• Calculation and comparison of dynamic property parameters for each algorithm using simulated EEG data.
• Quantitative analysis of estimation performance differences across the three algorithms.

Main Results:

• Observed significant differences in the estimation performance among the STFT, ADFT, and SPWD algorithms.
• The study identified varying accuracies and sensitivities in capturing dynamic brain communication patterns.
• Parameter analysis revealed distinct characteristics of each algorithm's dynamic property estimation.

Conclusions:

• The choice of time-variant coherence algorithm impacts the analysis of single-trial EEG brain communication.
• Algorithm performance differences necessitate careful consideration for specific neuroscience research applications.
• Further investigation into algorithm utilization for advanced brain function studies is warranted.