Physics-Guided Neural Networks for Intraventricular Vector Flow Mapping

• Published in: IEEE transactions on ultrasonics, ferroelectrics, and frequency control Vol. 71; no. 11; pp. 1377 - 1388
• Main Authors: Ling, Hang Jung, Bru, Salome, Puig, Julia, Vixege, Florian, Mendez, Simon, Nicoud, Franck, Courand, Pierre-Yves, Bernard, Olivier, Garcia, Damien
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.11.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE); Institute of Electrical and Electronics Engineers
• Subjects: Algorithms; Artificial Intelligence; Artificial neural networks; Biomarkers; Blood flow; Blood Flow Velocity - physiology; Boundary conditions; Cardiac flow; Cardiology and cardiovascular system; Color; color Doppler; Computational fluid dynamics; Computer Science; deep learning (DL); Doppler effect; echocardiography; Echocardiography, Doppler, Color - methods; Engineering Sciences; Flow mapping; Fluid dynamics; Fluids mechanics; Heart Ventricles - diagnostic imaging; Human health and pathology; Humans; Image acquisition; Image color analysis; Image enhancement; Image Processing, Computer-Assisted - methods; Image reconstruction; Imaging; In vivo methods and tests; Life Sciences; Machine Learning; Mechanics; Medical Imaging; Models, Cardiovascular; Neural networks; Neural Networks, Computer; Optimization; Physics; physics-guided neural networks (PGNNs); physics-informed neural networks (PINNs); Real time; Smoothing methods; ultrasound; vector flow imaging; Vectors; Physics-informed neural networks (PINNs); Vector flow imaging; Cardiac flow; Echocardiography; Color Doppler; Physics-guided neural networks (PGNNs); Deep learning (DL); Ultrasound
• ISSN: 0885-3010; 1525-8955; 2373-7840; 1525-8955
• DOI: 10.1109/TUFFC.2024.3411718

Intraventricular vector flow mapping (iVFM) seeks to enhance and quantify color Doppler in cardiac imaging. In this study, we propose novel alternatives to the traditional iVFM optimization scheme using physics-informed neural networks (PINNs) and a physics-guided nnU-Net-based supervised approach. When evaluated on simulated color Doppler images derived from a patient-specific computational fluid dynamics (CFD) model and in vivo Doppler acquisitions, both the approaches demonstrate comparable reconstruction performance to the original iVFM algorithm. The efficiency of PINNs is boosted through dual-stage optimization and pre-optimized weights. On the other hand, the nnU-Net method excels in generalizability and real-time capabilities. Notably, nnU-Net shows superior robustness on sparse and truncated Doppler data while maintaining independence from explicit boundary conditions. Overall, our results highlight the effectiveness of these methods in reconstructing intraventricular vector blood flow. The study also suggests potential applications of PINNs in ultrafast color Doppler imaging and the incorporation of fluid dynamics equations to derive biomarkers for cardiovascular diseases based on blood flow.