Graphics processing unit (GPU)-accelerated particle filter framework for positron emission tomography image...

Area of Science:

• Medical Imaging
• Computational Science
• Biophysics

Background:

• Positron emission tomography (PET) data are Poisson distributed due to random photon emissions and detections.
• Tracer kinetic modeling aims to link PET data with physiological parameters of tracer uptake and metabolism.
• Existing PET reconstruction often uses simplified models lacking physiological context.

Purpose of the Study:

• To develop a unified PET reconstruction framework incorporating both statistical and physiological models.
• To accelerate PET image reconstruction using graphics processing unit (GPU) parallel computing.
• To validate the proposed method with phantom and simulated data.

Main Methods:

• Formulation of organ activity distribution using tracer kinetic models.
• Development of observation equations for photon-counting measurements.
• Implementation of a GPU-accelerated reconstruction strategy using state-space evolution equations.

Main Results:

• The proposed method successfully integrates statistical and physiological constraints into a unified framework.
• GPU-based parallel computing significantly accelerates the reconstruction process.
• Experimental results demonstrate the method's effectiveness with various phantom datasets.

Conclusions:

• The developed GPU-accelerated strategy offers a powerful approach for PET image reconstruction.
• This method provides a practical solution for clinical applications by achieving efficient reconstruction times.
• Integrating physiological models enhances the accuracy and interpretability of PET imaging.