Data Assimilating B-splines for Model-based Regularization in Ultrasound Vector Flow Imaging.

Area of Science:

• Biomedical Engineering
• Fluid Dynamics
• Medical Imaging

Background:

• Ultrasound vector flow imaging (VFI) is crucial for non-invasive hemodynamic assessment.
• Accurate reconstruction of 4-D flow fields remains challenging due to noise and data limitations.
• Integrating fluid dynamics models can enhance VFI accuracy.

Purpose of the Study:

• To develop and validate a novel data assimilation framework for ultrasound VFI.
• To incorporate fluid dynamic models using tensor product B-splines for regularization.
• To investigate the benefits of this approach for improved flow field reconstruction.

Main Methods:

• A variational data assimilation method was developed using tensor product B-splines.
• The method solves high-dimensional regularization problems governed by the Navier-Stokes equations.
• Validation was performed in silico, in vitro (with PIV), and in vivo (compared to MRI).

Main Results:

• The proposed method demonstrated superior noise suppression and reduced RMSE compared to conventional techniques.
• Accurate reconstruction of flow features and gradient-based metrics was achieved in vitro.
• In vivo results showed good agreement with MRI, especially in high-velocity regions and dropout zones.

Conclusions:

• The data assimilation approach effectively reconstructs 4-D flow fields in ultrasound VFI.
• This method offers a promising solution for bedside clinical applications, balancing resolution and noise suppression.
• Leveraging physical models enhances the robustness of flow estimation in VFI.