Real-time ultrasound simulation using the GPU.

Area of Science:

• Medical Imaging
• Computational Ultrasound
• Computer Graphics

Background:

• Ultrasound simulators require real-time synthetic image generation for training.
• Current methods often use pre-calculated textures for speckle, limiting realism and dynamic object simulation.

Purpose of the Study:

• To develop a physically-based model for realistic speckle generation in ultrasound simulation.
• To enable real-time simulation of dynamic objects within ultrasound imaging.

Main Methods:

• A physical model based on the convolution of the ultrasound scanner's point spread function (PSF) with a scatterer distribution was employed.
• Scatterer selection and sampling were parallelized on a graphics processing unit (GPU).
• The model was validated using a cyst phantom and a movable needle.

Main Results:

• The proposed method achieved real-time speckle image generation (over 10 frames per second) on a standard GPU.
• Simulated ultrasound images demonstrated visual similarity to those from a reference method.
• The approach successfully imaged dynamic elements like a movable needle.

Conclusions:

• The GPU-accelerated physical model significantly enhances speckle realism and allows for dynamic object simulation in ultrasound simulators.
• This method offers a more advanced and realistic training tool for ultrasound image acquisition and interpretation.