Repository: https://github.com/isaac-for-healthcare/i4h-sensor-simulation

Raytracing Ultrasound Simulator#

A high-performance GPU-accelerated ultrasound simulator using NVIDIA OptiX raytracing.

The simulator implements the straight-ray approach of Bürger et al. (2013), IEEE Transactions on Medical Imaging: plausible, real-time B-mode images from labeled surface meshes, well suited to sim-to-real robotics perception and ML data generation. Wave effects such as diffraction, interference, and phase are not modeled — see capabilities and limitations before relying on it for quantitative acoustics or signal-processing work.

Features#

GPU acceleration with CUDA and NVIDIA OptiX
Python interface for ease of use
Real-time simulation capabilities
Support for curvilinear, linear, and phased array ultrasound probe simulation

Requirements#

Quick start#

Option 1: Using the I4H CLI (Recommended)#

The ./i4h CLI builds and runs inside a Docker container with all dependencies pre-installed.

# Launch the interactive web server (default mode)
./i4h run ultrasound-simulator

# Run the sphere-sweep demo (no mesh download needed)
./i4h run ultrasound-simulator sphere_sweep

# Run the liver-sweep demo
./i4h run ultrasound-simulator liver_sweep

# Run the performance benchmark
./i4h run ultrasound-simulator benchmark

# List available modes
./i4h modes ultrasound-simulator

# Launch an interactive shell inside the container
./i4h run-container ultrasound-simulator

Open your browser to http://0.0.0.0:8000 when running the server mode.

Note: With ./i4h, mesh-backed modes (server, liver_sweep, benchmark) use the default container mesh path (/opt/ultrasound-mesh) automatically. sphere_sweep does not use mesh assets.

Ultrasound Probe Simulation

Option 2: Using build_and_run.sh#

cd ultrasound-simulator
./build_and_run.sh examples/server.py

Option 3: Docker#

Instructions to build and run the examples in a docker environment can be found in the docs/docker_build.

Option 4: Bare-Metal Installation#

Instructions to build and run the examples on a bare-metal installation can be found in the docs/baremetal_build.

Start Simulating#

For a comprehensive guide on using the simulator, understanding its features, and exploring advanced topics, please refer to our documentation:

Getting Started Guide: A step-by-step tutorial for beginners.
Technical Guide: An in-depth look at the physics and implementation details.

Benchmark Results#

To reproduce these results, run python examples/benchmark.py.

Benchmark Results:
        Total frames: 200
        Average frame time: 0.0073 seconds
        Average FPS: 136.28
        Minimum FPS: 59.66
        Maximum FPS: 249.62
        Date: 2025-03-16 07:38:46

        System Information:
        GPU: NVIDIA RTX 6000 Ada Generation (48.0 GB, Driver: 565.57.01)
        CPU: AMD Ryzen Threadripper PRO 7975WX 32-Cores (64 cores)