Algorithm for Determination of Thresholds of Significant Coherence in Time-Frequency Analysis

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Quantitative coherence assessment is vital for analyzing biomedical signals.
• Applications include Transfer Function Analysis (TFA) of Cerebral Autoregulation (CA) and Coherent Hemodynamics Spectroscopy (CHS).
• Current methods require significant memory to store coherence distributions for threshold determination.

Purpose of the Study:

• To develop a memory-efficient algorithm for determining coherence thresholds.
• To tailor the algorithm for applications like TFA and CHS.
• To reduce computational memory usage in signal coherence analysis.

Main Methods:

• Utilized principles from data streaming algorithms for quantile approximation.
• Developed a novel algorithm to identify coherence thresholds from wavelet scaleograms.
• Focused on minimizing memory footprint compared to storing full coherence distributions.

Main Results:

• The algorithm successfully determines coherence thresholds with significantly reduced memory requirements.
• It is applicable to analyzing the coherence between Arterial Blood Pressure (ABP) and cerebral Blood Flow Velocity (BFV) or hemoglobin concentration.
• The method provides a practical solution for real-time or resource-constrained signal analysis.

Conclusions:

• The developed algorithm offers an efficient solution for calculating coherence thresholds in biomedical signal processing.
• It addresses the memory limitations of traditional methods, particularly for CA and CHS.
• This approach facilitates more accessible and scalable quantitative coherence analysis.