Dynamic 18F-FDG PET Images Simulation Using 4D-XCAT Phantom and Kinetic Modeling for Lesion Detectability Investigation and Scan Time Reduction Purpose

Abstract

Introduction: Simulation in Positron Emission Tomography (PET) studies is considered as an effective approach to test new mathematical methods for image processing and lesion detection. It’s an alternative way to overcome the drawback of obtaining a sufficient set of clinical images with known truth about the presence or absence of lesions. This work aimed to simulate, in a new and fast way, realistic dynamic 18F-FDG PET images for lesion detectability investigation and scan time reduction. Material and Methods: The 4D-XCAT phantom was utilized in this work. The three-compartment model was used to simulate the Time Activity Curves (TAC’s) of 18F-FDG. The arterial input function of 18F-FDG was modeled using a parametric function. The TAC’s of 11 tissues defined in the 4D-XCAT phantom were simulated. The activity values were calculated from the TAC’s considering a real 18F-FDG dynamic PET acquisition protocol. These activity values were assigned to each voxel of 4D-XCAT to produce 28 activity maps. The GE Discovery PET/CT 710 scanner, modeled in the STIR platform, was used to generate the sinograms. OSMAPOSL Algorithm was considered to reconstruct dynamic 18F-FDG PET images. Results: Realistic dynamic 18F-FDG PET images were generated. The qualitative and quantitative comparison showed a good agreement between the 4D-XCAT phantom images before and after the reconstruction procedure. The computation time of the reconstruction procedure was 8.76 min/frame. Conclusion: The present study was found to be a promising and realistic approach to dynamic PET dPET imaging optimization in terms of scanning time reduction and lesion detectability amelioration.

Keywords

• computer simulation positron
• emission tomography imaging phantom image reconstruction