Ultrasound simulation with deformable and patient-specific scatterer maps.

Area of Science:

• Medical Imaging
• Computational Ultrasound Simulation

Background:

• Current ray-tracing simulations struggle to model realistic ultrasound (US) texture and dynamic tissue deformations.
• Manual parameterization of scatterer representations is complex and does not account for volumetric changes during tissue compression.

Purpose of the Study:

• To enhance the realism of ray-tracing-based ultrasound simulations by incorporating dynamic speckle patterns that adapt to tissue deformation.
• To utilize brightness mode (B-mode) estimated scatterer maps for improved ray tracing.

Main Methods:

• Simulated US texture deformations in the scatterer domain using back-projection of ray segments.
• Estimated scatterer maps from in vivo images via a pretrained generative adversarial network.
• Integrated dynamic speckle patterns that change with tissue deformation during simulation.

Main Results:

• Demonstrated a novel scatterer estimation and runtime background fusion method on simulated transvaginal US scans.
• Achieved interactive frame rates of 28 frames per second for modeling deformations.
• Quantitative and qualitative evaluations showed improved realism compared to existing methods.

Conclusions:

• Physically consistent incorporation of deformations in the scatterer domain enhances simulation realism.
• Seamlessly compounds anatomical content and dynamic deformations using a unified imaging model.
• Enables more accurate and visually convincing ultrasound simulations.