Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2019 Mar;47(3):714-730.
doi: 10.1007/s10439-018-02191-z. Epub 2019 Jan 3.

Data-Augmented Modeling of Intracranial Pressure

Affiliations

Data-Augmented Modeling of Intracranial Pressure

Jian-Xun Wang et al. Ann Biomed Eng. 2019 Mar.

Abstract

Precise management of patients with cerebral diseases often requires intracranial pressure (ICP) monitoring, which is highly invasive and requires a specialized ICU setting. The ability to noninvasively estimate ICP is highly compelling as an alternative to, or screening for, invasive ICP measurement. Most existing approaches for noninvasive ICP estimation aim to build a regression function that maps noninvasive measurements to an ICP estimate using statistical learning techniques. These data-based approaches have met limited success, likely because the amount of training data needed is onerous for this complex applications. In this work, we discuss an alternative strategy that aims to better utilize noninvasive measurement data by leveraging mechanistic understanding of physiology. Specifically, we developed a Bayesian framework that combines a multiscale model of intracranial physiology with noninvasive measurements of cerebral blood flow using transcranial Doppler. Virtual experiments with synthetic data are conducted to verify and analyze the proposed framework. A preliminary clinical application study on two patients is also performed in which we demonstrate the ability of this method to improve ICP prediction.

Keywords: Cerebrovascular dynamics; Data assimilation; Patient-specific modeling; Transcranial Doppler.

PubMed Disclaimer

Figures

Figure 1:
Figure 1:
Schematic of the proposed data-augmented, theory-based framework for ICP dynamics. By assimilating noninvasively measurable data (e.g., CBFV and/or ABP at certain vessels) into the theory-based physiological model, predictions of the unobservable states (e.g., ICP) can be significantly improved.
Figure 2:
Figure 2:
Schematic of the multiscale cerebrovascular model coupling (a) a distributed 1D propagation network model for major systemic arteries and (b) a lumped parameter (LP) network for intracranial dynamics. Outflow in the 1D portion marked with open circles are coupled with the LP network in (right), and boundaries marked with closed circles are coupled to 3-element Windkessel models. The bounding box represents intracranial space. Adapted from
Figure 3:
Figure 3:
Schematic of the iterative ensemble Kalman method. (1) Initial ensemble is obtained by sampling the prior parameter space and (2) is propagated through the forward intracranial model. (3) The propagated state will be updated by assimilating TCD measurement data by Bayesian analysis. Steps (2) and (3) will be conducted iteratively until reaching the statistical convergence.
Figure 4:
Figure 4:
Iteration histories of unknown parameters (i.e., target flow rates qn,n=1,6) by assimilating noise-free synthetic TCD data. The prior ensemble of each parameter is biased from the respect truth.
Figure 5:
Figure 5:
Comparison of (a) prior ICP prediction and posterior ICP predictions following (b) noise-free, two MCAs, (c) noisy, two MCAs, (d) noise-free, right MCA and (e) noisy, right MCA CBFV data assimilations.
Figure 6:
Figure 6:
Comparison of prior and posterior predictions of CBFV at right ACA following noisy synthetic data assimilation. In (b) CBFV data at two MCAs with 10% Gaussian noises are assimilated; In (c) Both CBFV and ABP data at two MCAs with 10% Gaussian noises are assimilated.
Figure 7:
Figure 7:
TCD CBFV data. (a) Raw TCD-based CBFV signals at right MCA of the patient P1 over 260 cardiac cycles. (b-c) Aggregated CBFV pulses at the right MCA and its ensemble averaged for (b) patient P1 and (c) patient P2.
Figure 8:
Figure 8:
CBFV predictions (a-d) and ICP predictions (e-h) following assimilation of TCD CBFV measurements.

Similar articles

Cited by

References

    1. Andrews PJ, Citerio G, Longhi L, Polderman K, Sahuquillo J, Vajkoczy P, Care N-I, E. M. N. S. of the European Society of Intensive Care Medicine et al. NICEM consensus on neurological monitoring in acute neurological disease. Intensive care medicine 34:1362–1370, 2008. - PubMed
    1. Arnold A, Battista C, Bia D, German YZ, Armentano RL, Tran H, and Olufsen MS. Uncertainty quantification in a patient-specific one-dimensional arterial network model: EnKF-based inflow estimator. Journal of Verification, Validation and Uncertainty Quantification 2:011002, 2017. - PMC - PubMed
    1. Arroyo-Palacios J, Rudz M, Fidler R, Smith W, Ko N, Park S, Bai Y, and Hu X. Characterization of shape differences among ICP pulses predicts outcome of external ventricular drainage weaning trial. Neurocritical care 25:424–433, 2016. - PMC - PubMed
    1. Asiedu DP, Lee K-J, Mills G, and Kaufmann EE. A review of non-invasive methods of monitoring intracranial pressure. Journal of Neurology Research 4:1–6, 2014.
    1. Bertoglio C, Moireau P, and Gerbeau J-F. Sequential parameter estimation for fluid–structure problems: Application to hemodynamics. International Journal for Numerical Methods in Biomedical Engineering 28:434–455, 2012. - PubMed

LinkOut - more resources