Merging computational fluid dynamics and 4D Flow MRI using proper orthogonal decomposition and ridge regression

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Fluid Dynamics

Background:

• Time-resolved phase-contrast MRI (4D Flow MRI) measures blood velocity but suffers from noise and resolution limits.
• Existing methods struggle to accurately reconstruct detailed flow fields from 4D Flow MRI data.

Purpose of the Study:

• To develop a novel method combining 4D Flow MRI with computational fluid dynamics (CFD).
• To address limitations of 4D Flow MRI, such as noise and resolution limits, for high-fidelity flow field reconstruction.
• To improve the accuracy of clinically relevant hemodynamic parameters derived from MRI.

Main Methods:

• Proper Orthogonal Decomposition (POD) was used to create orthonormal bases for flow equations at both low (MRI) and high (CFD) resolutions.
• In vivo 4D Flow MRI data was projected onto low-resolution basis vectors for de-noising.
• Ridge regression was employed to reconstruct high-resolution, de-noised, and divergence-free flow fields.

Main Results:

• The POD-based method successfully reconstructed de-noised, high-resolution flow fields.
• Compared to state-of-the-art methods using a cerebral aneurysm phantom, POD preserved small flow structures and eliminated noise at MRI resolution.
• The method revealed flow details at CFD mesh resolution not visible at the original MRI resolution.

Conclusions:

• The novel 4D Flow MRI and CFD merging method effectively overcomes noise and resolution limitations of standard MRI.
• This approach enhances the accuracy of crucial hemodynamic parameters like pressure gradients and wall shear stress.
• The technique holds significant potential for improving cardiovascular research and clinical diagnostics.