Fast GPU-based Monte Carlo code for SPECT/CT reconstructions generates improved ¹⁷⁷Lu images.

Area of Science:

• Nuclear Medicine
• Medical Imaging
• Computational Science

Background:

• Full Monte Carlo (MC) SPECT reconstructions offer potential for image correction but are time-consuming.
• Developing faster MC codes is crucial for clinical application of advanced SPECT/CT imaging.

Purpose of the Study:

• To develop a highly parallel MC code for rapid Ordered Subset Expectation Maximization (OSEM) reconstructions of SPECT/CT images.
• To accelerate MC-based SPECT/CT image reconstruction using Graphics Processing Units (GPUs).

Main Methods:

• A parallel MC code was developed using Compute Unified Device Architecture (CUDA) on a multi-GPU system.
• The code simulated photon emissions, incorporating attenuation, scattering, and detector physics.
• MC-OSEM was applied to phantom and 177-Lu SPECT/CT images, comparing performance with standard AC-OSEM and RRC-OSEM.

Main Results:

• The MC code achieved a performance of 3200 million photons/s.
• MC-based OSEM reconstruction was completed in approximately 3 minutes per patient scan.
• Activity recovery in phantom studies showed significant improvement (88% for largest sphere) compared to AC-OSEM (66%) and RRC-OSEM (79%).

Conclusions:

• GPU-accelerated MC code enables rapid SPECT/CT reconstruction within minutes.
• Reconstructed clinical images demonstrated enhanced resolution and contrast, particularly for 177-Lu-DOTATATE treatments.