FUX-Sim: Implementation of a fast universal simulation/reconstruction framework for X-ray systems.

Area of Science:

• Medical Imaging
• Computational Science
• Computer Engineering

Background:

• Digital X-ray detectors and advanced reconstruction algorithms enable 3D imaging in radiology.
• Iterative reconstruction algorithms for non-standard protocols are computationally intensive.
• Computer simulations are crucial for developing and validating new X-ray systems.

Purpose of the Study:

• To develop a flexible and fast X-ray simulation/reconstruction framework.
• To address computational burdens and hardware/software optimization challenges.
• To support exploration of novel scanning geometries and multiple programming models.

Main Methods:

• Developed FUX-Sim, a novel, modular, layered X-ray simulation/reconstruction framework.
• Implemented parallel algorithms for Graphics Processing Units (GPUs) using CUDA and OpenCL, and multi-core Central Processing Units (CPUs).
• Optimized for hardware and software efficiency to handle large data volumes and flexible geometries.

Main Results:

• FUX-Sim demonstrates high performance, particularly with the CUDA programming model (5x faster than state-of-the-art).
• The framework supports a wide range of hardware platforms through CPU and OpenCL implementations.
• Achieved optimized performance for diverse system configurations and hardware.

Conclusions:

• FUX-Sim provides a flexible and efficient solution for X-ray simulation and reconstruction.
• The framework facilitates the development of advanced 3D radiology systems by enabling rapid performance checking.
• Compatibility with multiple programming models and platforms ensures adaptability for future hardware evolution.