Optimized regional myocardial blood flow abnormality detection in dynamic Rb-82 PET

Objectives The goal is to optimize the dynamic Rb-82 PET reconstruction/acquisition protocol for maximum detection of reduced regional tracer transport from blood pool to myocardial tissue.

Methods Stress time activity curves (TACs) of blood pool and myocardium up to 2 min acquisition were extracted from dynamic Rb-82 PET images of 5 normal patients. They were smoothed and averaged to determine typical Rb-82 biodistribution in a normal patient. A myocardial TAC corresponding to reduced regional myocardial blood flow (MBF) was generated from the normal blood TAC and designed kinetic rate constants. The normal and reduced MBF sets of TACs were used to create two 4D NCAT phantom sets. Combined SimSET-GATE Monte-Carlo simulation was used to generate nearly-noise-free (NNF) PET data that were scaled to clinical count level before Poisson noise was added. Simulated datasets from different acquisition schemes were reconstructed using the OS-EM algorithm with post-filtering. Kinetic rate constants were extracted from the NNF and noisy reconstructed images using a two-compartment model. The reconstruction and post-filtering parameters were optimized for accurate kinetic parameter estimation from the NNF images. The regional AUC values were obtained from noisy images to classify normal and reduced MBF hearts.

Results With optimized image reconstruction and post-filtering protocol, the 10*12sec acquisition scheme provides the highest regional AUC values for distinguishing normal and reduced MBF hearts through compartmental analyses.

Conclusions Realistic simulation based on clinical data was used to optimize dynamic Rb-82 PET reconstruction/acquisition protocol. The optimized protocol was determined for maximum detection of reduced regional MBF using a two-compartment kinetic model.

Research Support NIH R01-EB000168