Introducing improved voxel navigation and fictitious interaction tracking in GATE for enhanced efficiency

Area of Science:

• Medical Physics
• Computational Imaging
• Nuclear Medicine

Background:

• Geant4 Application for Emission Tomography (GATE) is a versatile tool for emission tomography simulations.
• Its computational performance is limited, particularly with voxelized phantoms, due to general particle tracking algorithms.

Purpose of the Study:

• To enhance the computational performance of GATE simulations.
• To reduce the time dedicated to particle tracking within phantoms.

Main Methods:

• Implemented and evaluated Geant4's 'regular navigation algorithm'.
• Applied fictitious interaction tracking (Woodcock tracking) for photon simulation.
• Simulated a clinical head-and-neck [(18)F]FDG PET/CT scan using the Allegro/GEMINI GXL PET/CT scanner.

Main Results:

• Achieved significant speed-up factors ranging from approximately 2.7 to 27.6.
• New methods demonstrated minimal impact on simulation accuracy, with less than 0.5% differences in detected coincidences, trues, scatter, and random fractions.
• Performance gains were observed across various phantom voxel sizes.

Conclusions:

• The proposed tracking methods substantially accelerate GATE simulations for emission tomography.
• These optimizations improve computational efficiency for complex simulations, such as PET/CT scans.
• The enhanced methods offer a practical solution for reducing simulation times in medical imaging research.