Scatter Correction based on GPU-accelerated Full Monte Carlo Simulation for Brain PET/MRI
08 July 2019
Scatter in PET is one of the major sources of degradation and bias in quantification, and, therefore, accurate scatter correction is essential for qualitative and quantitative PET imaging. The aim of this work was to develop a full Monte Carlo simulation based on GPU-acceleration (graphics processing unit) for PET applications. This allows physically accurate scatter estimations within a reasonable computational time, compatible with clinical requirements. The presented results confirm the potential of this approach. The new proposed method outperforms the standard SSS (single scatter simulation) method with respect to recovery coefficient and correction accuracy in phantom measurements. Finally, the representative patient examples demonstrate the applicability of the method in the image reconstruction of real patient data. The differences in TBRs (tumour to brain ratio) and tumour volumes demonstrate the clinical relevance of applying accurate scatter corrections.