Assessing chance in neuro-vascular interactions.

Area of Science:

• Neuroscience
• Physiology
• Biophysics

Background:

• Interpreting correlations between neuronal activity and hemodynamic signals is challenging due to strong autocorrelation.
• Standard statistical tests can overestimate neurovascular coupling during rest by underestimating false positives.

Purpose of the Study:

• To develop and validate a surrogate-based statistical framework for calibrating inference in autocorrelated physiological signals.
• To accurately assess neurovascular coupling by distinguishing genuine interactions from random associations in resting-state data.

Main Methods:

• Utilized simultaneous recordings of cortical oxygen tension (PO₂), single-unit firing, and local field potentials (LFP) in awake rabbits.
• Applied amplitude-adjusted Fourier surrogates to generate null distributions that preserve temporal structure while removing cross-dependence.
• Integrated lag optimization, multiple comparison controls, and population-level inference scaling.

Main Results:

• Found significant PO₂ correlations with delta-band LFP and a subset of single neurons, exceeding chance levels under surrogate testing.
• Correlations with other LFP bands were not significant, and aggregated neuronal activity (multi-unit signals) did not predict PO₂.
• Identified that small, synchronized neuronal subpopulations, rather than global activity, produced robust associations with PO₂.

Conclusions:

• Refined understanding of resting-state neurovascular coupling, demonstrating that apparent broad correlations resolve into selective and reproducible effects with calibrated testing.
• The surrogate-based inference framework effectively prevents misinterpretation of autocorrelated data.
• The proposed method offers a generalizable approach for time-series analysis in various scientific domains, including electrophysiology and neuroimaging.