Dataset Information

Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations.

ABSTRACT: For centuries, flow visualization has been the art of making fluid motion visible in physical and biological systems. Although such flow patterns can be, in principle, described by the Navier-Stokes equations, extracting the velocity and pressure fields directly from the images is challenging. We addressed this problem by developing hidden fluid mechanics (HFM), a physics-informed deep-learning framework capable of encoding the Navier-Stokes equations into the neural networks while being agnostic to the geometry or the initial and boundary conditions. We demonstrate HFM for several physical and biomedical problems by extracting quantitative information for which direct measurements may not be possible. HFM is robust to low resolution and substantial noise in the observation data, which is important for potential applications.

SUBMITTER: Raissi M

PROVIDER: S-EPMC7219083 | biostudies-literature | 2020 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations.

Raissi Maziar M Yazdani Alireza A Karniadakis George Em GE

Science (New York, N.Y.) 20200130 6481

For centuries, flow visualization has been the art of making fluid motion visible in physical and biological systems. Although such flow patterns can be, in principle, described by the Navier-Stokes equations, extracting the velocity and pressure fields directly from the images is challenging. We addressed this problem by developing hidden fluid mechanics (HFM), a physics-informed deep-learning framework capable of encoding the Navier-Stokes equations into the neural networks while being agnosti ...[more]

PMID: 32001523

Similar Datasets

Project description:BackgroundGlomerular hyperfiltration resulting from an elevated intraglomerular pressure (Pglom) is an important cause of CKD, but there is no feasible method to directly assess Pglom in humans. We developed a model to estimate Pglom in patients from combined renal arterial pressure and flow measurements.MethodsWe performed hemodynamic measurements in 34 patients undergoing renal or cardiac angiography under baseline conditions and during hyperemia induced by intrarenal dopamine infusion (30 μg/kg). For each participant during baseline and hyperemia, we fitted an adapted three-element Windkessel model that consisted of characteristic impedance, compliance, afferent resistance, and Pglom.ResultsWe successfully analyzed data from 28 (82%) patients. Median age was 58 years (IQR, 52-65), median eGFR was 95 ml/min per 1.73 m2 (IQR, 74-100) using the CKD-EPI formula, 30% had microalbuminuria, and 32% had diabetes. The model showed a mean Pglom of 48.0 mm Hg (SD=10.1) at baseline. Under hyperemia, flow increased by 88% (95% CI, 68% to 111%). This resulted in a 165% (95% CI, 79% to 294%) increase in afferent compliance and a 13.1-mm Hg (95% CI, 10.0 to 16.3) decrease in Pglom. In multiple linear regression analysis, diabetes (coefficient, 10.1; 95% CI, 5.1 to 15.1), BMI (0.99 per kg/m2; 95% CI, 0.38 to 1.59), and renal perfusion pressure (0.42 per mm Hg; 95% CI, 0.25 to 0.59) were significantly positively associated with baseline Pglom.ConclusionsWe constructed a model on the basis of proximal renal arterial pressure and flow velocity measurements that provides an overall estimate of glomerular pressure and afferent and efferent resistance in humans. The model provides a novel research technique to evaluate the hemodynamics of CKD on the basis of direct pressure and flow measurements.Clinical trial registry name and registration numberFunctional HEmodynamics in patients with and without Renal Artery stenosis (HERA), NL40795.018.12 at the Dutch national trial registry (toetsingonline.nl).

Dataset Information

Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations.

Publications

Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets