GPU-Based Simulation of Echocardiography Volumes Using Quantitative Fiber-Angle-to-Backscatter Measurements

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Computational Science

Background:

• Ultrasound backscatter intensity correlates with cardiac fiber orientation relative to the insonification beam.
• Accurate 3D visualization of cardiac muscle fiber architecture is crucial for understanding heart function and disease.

Purpose of the Study:

• To develop and validate a GPU-based simulation method for generating 3D echocardiographic images from an established angle-to-backscatter relationship.
• To assess the simulation's ability to produce view-dependent speckle and accurately map cardiac fiber orientation.

Main Methods:

• A GPU-based simulation approach was developed using an empirically derived angle-to-backscatter relationship.
• Simulations were performed on rotating fiber phantoms, DT-MRI data of a canine heart, and parametric left ventricle models.
• Image data were acquired using simulated 2D matrix array and 1D phased array probes.

Main Results:

• The simulation accurately reflected the angle-to-backscatter relationship in phantom studies.
• Simulated echocardiograms from DT-MRI data demonstrated view-dependent speckle.
• Processed fiber orientation maps from simulated 3D volumes showed good agreement with ground truth, with an average acute angle error (AAE) under 10 degrees for both probe types.

Conclusions:

• The developed GPU-based simulation method is efficient and accurately reconstructs cardiac fiber orientation from ultrasound data.
• This approach offers a novel and fast method for ultrasound-based cardiac fiber imaging, with potential clinical applications.