Related Experiment Videos
Linear and nonlinear analysis of human dynamic cerebral autoregulation
R B Panerai1, S L Dawson, J F Potter
1Division of Medical Physics, University of Leicester, Leicester Royal Infirmary, Leicester LE1 5WW, United Kingdom. rp9@le.ac.uk
The American Journal of Physiology
|September 14, 1999
Summary
This study explored the relationship between arterial blood pressure and cerebral blood flow velocity. While linear models offer some accuracy, a nonlinear component significantly improves modeling of cerebral autoregulation in humans.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Physiology
Background:
- Cerebral autoregulation is crucial for maintaining stable brain blood flow.
- Understanding the dynamic relationship between arterial blood pressure (ABP) and cerebral blood flow velocity (CBFV) is vital for assessing cerebrovascular health.
- Previous models primarily focused on linear dynamics, potentially overlooking nonlinear influences.
Purpose of the Study:
- To investigate the linear and nonlinear dynamic relationships between systemic arterial blood pressure (ABP) and cerebral blood flow velocity (CBFV) in humans.
- To compare the predictive accuracy of linear versus nonlinear models in dynamic cerebral autoregulation.
- To determine the contribution of nonlinear components to modeling CBFV responses.
Main Methods:
- Employed time- and frequency-domain analysis for linear dynamics.
- Utilized a nonlinear moving-average approach with Volterra-Wiener kernels.
- Measured ABP using Finapres and CBFV via Doppler ultrasound in 47 healthy subjects under resting and induced hypotension conditions.
Main Results:
- Linear models (Fourier transfer function, Tiecks model, linear Volterra-Wiener kernel) showed significant correlations (r ≈ 0.52-0.67) between reconstructed and recorded CBFV.
- A second-order nonlinear model (linear plus quadratic kernels) significantly improved accuracy (r = 0.82 +/- 0.08) for the same data segment.
- Linear methods performed equivalently when predicting different data segments or transient responses, whereas the nonlinear model's predictive power dropped significantly (r ≈ 0.21-0.26).
Conclusions:
- Dynamic cerebral autoregulation in humans can be modeled using linear methods.
- A second-order nonlinear component significantly enhances model accuracy for specific data segments during autoregulation assessment.
- The nonlinear component's predictive capability does not automatically extend to different data segments or transient physiological changes.