Model-based inference from microvascular measurements: Combining experimental measurements and model predictions...

Area of Science:

• Physiology
• Biophysics
• Computational Biology

Background:

• In vivo microcirculation imaging and network modeling are crucial for understanding microvascular function.
• High-throughput imaging generates vast data, necessitating methods to extract key physiological indices.
• Accurate estimation of these indices requires network analysis robust to measurement and model errors.

Purpose of the Study:

• To propose a Bayesian probabilistic data analysis framework for integrating microvascular imaging data and network model simulations.
• To develop a statistically coherent method for analyzing microvascular function.
• To address uncertainty inherent in experimental measurements and model errors.

Main Methods:

• Developed a Bayesian probabilistic data analysis framework.
• Integrated experimental measurements with network model simulations.
• Handled noisy measurements and provided posterior distributions for parameters and indices.

Main Results:

• Applied the framework to rat mesentery and mouse brain cortex microvascular networks.
• Inferred distributions for over 500 unknown pressure and hematocrit boundary conditions.
• Demonstrated robustness of model predictions even when measurements were omitted during calibration.

Conclusions:

• The Bayesian probabilistic approach offers a statistically coherent method for analyzing microvascular data.
• This framework is suitable for optimizing data acquisition in microvascular imaging studies.
• It aids in analyzing and reporting large datasets from microvascular research.