Object independent scatter sensitivities for PET, applied to scatter estimation through fast Monte Carlo simulation

Area of Science:

• Medical Imaging
• Nuclear Physics
• Computational Science

Background:

• Quantitative bias in Positron Emission Tomography (PET) is significantly influenced by scattered coincidences.
• Current scatter estimation methods often use simplified scanner models, neglecting crucial detector physics.
• Accurate compensation requires estimating scattered coincidences per line-of-response and time-of-flight bin.

Purpose of the Study:

• To develop and validate an improved method for scatter estimation in PET by incorporating detector physics.
• To assess the impact of this method on quantitative accuracy under various simulation conditions, including low-count statistics.
• To demonstrate the feasibility of using fast Monte Carlo simulations for robust scatter estimation in clinical settings.

Main Methods:

• Integrated a 5D single-photon detection probability lookup table (energy, angle, location) into a Monte Carlo simulator (MCGPU-PET).
• Simulated scatter using MCGPU-PET and applied it to phantom data from a simulated GE Signa PET/MR scanner in GATE.
• Evaluated three scenarios: high-count simulations, limited-count simulations, and joint activity-scatter estimation under low-count conditions.

Main Results:

• Scatter-compensated reconstructions achieved less than 1% global bias in high-count simulations.
• Gaussian smoothing restored accuracy in limited-count simulations where noisy scatter estimates initially caused bias.
• Joint estimation maintained less than 1% global bias even under low-count acquisition conditions.

Conclusions:

• The proposed scatter sensitivity modeling enhances existing clinical simulators by including detector physics.
• Fast Monte Carlo simulation is feasible for accurate and robust scatter estimation in real PET scans.
• This approach offers potential for improved quantitative accuracy and noise robustness in PET imaging.