NVIDIA OptiX ray-tracing engine as a new tool for modelling medical imaging systems

Area of Science:

• Medical Physics
• Computational Imaging
• Scientific Computing

Background:

• Monte Carlo simulations offer accurate modeling of X- and gamma radiation but are computationally intensive.
• Ray tracing provides a faster alternative by integrating interaction probabilities along paths.
• Both methods are crucial for simulating medical imaging systems and in iterative tomographic reconstruction algorithms.

Purpose of the Study:

• To introduce and evaluate the NVIDIA OptiX ray-tracing engine as a tool for medical imaging simulations.
• To demonstrate the engine's capability in implementing physical interactions and recording relevant quantities.
• To develop a rapid Monte Carlo simulation code for Single Photon Emission Computed Tomography (SPECT) and a ray-tracing projector.

Main Methods:

• Application of the NVIDIA OptiX engine for X- and gamma ray-tracing.
• Implementation of physical interactions with various object representations (pixel, mesh, NURBS).
• Development of GPU-accelerated code for Monte Carlo simulations and ray-tracing projectors, tested on heterogeneous multi-GPU systems.

Main Results:

• The OptiX engine enables efficient, scalable, and optimized GPU code for ray tracing and Monte Carlo simulations.
• Computation time for Monte Carlo simulations was reduced from days to minutes.
• The implemented SPECT simulation and projector showed comparable results to the GATE package.

Conclusions:

• NVIDIA OptiX is a powerful and efficient tool for accelerating ray-tracing and Monte Carlo simulations in medical imaging.
• The developed code facilitates rapid simulations for SPECT imaging and iterative reconstruction algorithms.
• Massive parallelization on GPUs significantly enhances computational performance, making complex simulations more accessible.