We formulate a physics-informed compressed sensing (PICS) method for the reconstruction of velocity fields from noisy and sparse phase-contrast magnetic resonance signals. The method solves an inverse Navier-Stokes boundary value problem, which permits us to jointly reconstruct and segment the velocity field, and at the same time infer hidden quantities such as the hydrodynamic pressure and the wall shear stress. Using a Bayesian framework, we regularize the problem by introducing a priori information about the unknown parameters in the form of Gaussian random fields. This prior information is updated using the Navier-Stokes problem, an energy-based segmentation functional, and by requiring that the reconstruction is consistent with the k-space signals. We create an algorithm that solves this inverse problem, and test it for noisy and sparse k-space signals of the flow through a converging nozzle. We find that the method is capable of reconstructing and segmenting the velocity fields from sparsely-sampled (15% k-space coverage), low (~10) signal-to-noise ratio (SNR) signals, and that the reconstructed velocity field compares well with that derived from fully-sampled (100% k-space coverage) high (>40) SNR signals of the same flow.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.1109/TIP.2022.3228172 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
Article Synopsis
- Post-stroke upper limb dysfunction significantly reduces patients' quality of life, and using sEMG signals to predict motion intentions can improve rehabilitation outcomes by adjusting robotic assistance levels.
- This paper introduces four deep learning models focusing on muscle synergy (MS) feature extraction, with a notable model utilizing 3D Convolutional Neural Networks (3DCNN) that processes sEMG data from an anatomical perspective for enhanced accuracy.
- Experimental results show the 3DCNN model outperforms other models in predicting wrist motion, achieving impressive accuracy and efficiency metrics, and demonstrates superiority over traditional musculoskeletal and deep learning models.
Discovering 3D hidden elasticity in isotropic and transversely isotropic materials with physics-informed UNets.
Acta Biomater
August 2024
Department of Biomedical Engineering, University of Arizona College of Engineering, Tucson, AZ, USA; Department of Aerospace and Mechanical Engineering, University of Arizona College of Engineering, Tucson, AZ, USA; Department of Mechanical Engineering, University of California Riverside, Riverside, CA, USA. Electronic address:
Three-dimensional variation in structural components or fiber alignments results in complex mechanical property distribution in tissues and biomaterials. In this paper, we use a physics-informed UNet-based neural network model (El-UNet) to discover the three-dimensional (3D) internal composition and space-dependent material properties of heterogeneous isotropic and transversely isotropic materials without a priori knowledge of the composition. We then show the capabilities of El-UNet by validating against data obtained from finite-element simulations of two soft tissues, namely, brain tissue and articular cartilage, under various loading conditions.
View Article and Find Full Text PDFUnsupervised Denoising and Super-Resolution of Vascular Flow Data by Physics-Informed Machine Learning.
J Biomech Eng
September 2024
Department of Mechanical Engineering, University of California, Berkeley, CA 94501.
We present an unsupervised deep learning method to perform flow denoising and super-resolution without high-resolution labels. We demonstrate the ability of a single model to reconstruct three-dimensional stenosis and aneurysm flows, with varying geometries, orientations, and boundary conditions. Ground truth data was generated using computational fluid dynamics, and then corrupted with multiplicative Gaussian noise.
View Article and Find Full Text PDFContinuous and discontinuous compressible flows in a converging-diverging channel solved by physics-informed neural networks without exogenous data.
Sci Rep
February 2024
State Key Laboratory of Fluid Power and Mechatronic Systems, Department of Engineering Mechanics, Zhejiang University, Hangzhou, 310027, China.
Physics-informed neural networks (PINNs) are employed to solve the classical compressible flow problem in a converging-diverging nozzle. This problem represents a typical example described by the Euler equations, a thorough understanding of which serves as a guide for solving more general compressible flows. Given a geometry of the channel, analytical solutions for the steady states do indeed exist, and they depend on the ratio between the back pressure of the outlet and the stagnation pressure of the inlet.
View Article and Find Full Text PDFLagrangian large eddy simulations via physics-informed machine learning.
Proc Natl Acad Sci U S A
August 2023
Graduate Interdisciplinary Program in Applied Mathematics and Department of Mathematics, University of Arizona, Tucson, AZ 85721.
High-Reynolds number homogeneous isotropic turbulence (HIT) is fully described within the Navier-Stokes (NS) equations, which are notoriously difficult to solve numerically. Engineers, interested primarily in describing turbulence at a reduced range of resolved scales, have designed heuristics, known as large eddy simulation (LES). LES is described in terms of the temporally evolving Eulerian velocity field defined over a spatial grid with the mean-spacing correspondent to the resolved scale.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!