Effective brain connectivity for fNIRS data analysis based on multi-delays symbolic phase transfer entropy

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Effective connectivity (EC) calculations for fNIRS data are often compromised by noise, leading to inaccurate results.
• Existing EC methods fail to account for diverse neurotransmission delays between brain regions, using fixed coefficients instead.

Purpose of the Study:

• To introduce a novel method, delay symbolic phase transfer entropy (dSPTE), to enhance EC estimation for fNIRS data.
• To address limitations of existing EC methods, specifically noise sensitivity and the handling of variable inter-region delays.

Main Methods:

• Developed dSPTE by integrating phase information from Hilbert transforms and state-space reconstruction with symbolic techniques (neural-gas algorithm).
• Applied a multi-time delay scale approach to accurately capture varying neurotransmission delays between brain regions.
• Validated dSPTE using linear AR, nonlinear, and multivariate hybrid models to simulate its performance.

Main Results:

• dSPTE achieved the highest accuracy (74.27%) and 100% specificity, indicating no false connectivity.
• The method demonstrated superior noise robustness and accurate identification of connectivity, even with significant delays.
• Analysis of fNIRS data during a finger-tapping task showed significantly increased EC strength in the task state compared to the resting state.

Conclusions:

• The dSPTE method offers a promising advancement for measuring EC in fNIRS data.
• This is the first application of phase information transfer entropy combined with symbolic processing for fNIRS analysis.
• dSPTE is confirmed to be noise-robust and suitable for analyzing complex brain networks with varying coupling delays.