Evaluation of Joint Reconstruction of Activity and Attenuation in Time-of-Flight PET
Jingnan Wang,Tong Wang,Wenjia Zhu,Yue Zhang,Nan Li,Hui Zhang,Li Huo
2019-01-01
Journal of Nuclear Medicine
Abstract:243 Objectives: Attenuation correction is crucial for quantitative PET imaging. As a standard method, CT images are used to derive attenuation correction factors. However, under some circumstances, attenuation correction based on CT images may bring false positive readings or artifacts due to high density implants in vivo or mismatch between PET and CT caused by motion. Recently, time-of-flight (TOF) PET has been brought into commercial use. In general, TOF information helps to improve consistency in the reconstruction and makes it less prone to inconsistencies between emission data and corrections. In this study, the maximum likelihood activity and attenuation estimation (MLAA) algorithm is used in the joint reconstruction of the emission and attenuation from TOF PET projections. Our study aims to evaluate the feasibility of MLAA based attenuation correction on both image quality and quantitative analysis of 18F-FDG PET imaging. Methods: To make the comparison between MLAA-based attention correction and traditional CT-based attenuation in PET image reconstructions, both phantom and preliminary patient studies were carried out. A uniform 68Ge cylindrical phantom with an activity of 25.9 MBq (0.7 mCi) was scanned and reconstructed using OSEM with CT-based attenuation correction and the MLAA joint reconstruction algorithm respectively. SUVs were calculated and compared between the two methods. In addition, PET images from 15 patients who underwent whole-body 18F-FDG PET/CT were reconstructed using the two reconstruction methods. Image qualities were visually compared. SUVmax and SUVmean values in liver, spleen, lungs, blood pool, bone, muscle and heart were quantitative analyzed. Results: Visual inspections of 68Ge cylinder scan images of the MLAA-based attenuation correction suggested that satisfactory image quality could be achieved without significant artifacts, with 4.55% difference in SUVmean (MLAA-based SUVmean 0.92 vs. CT-based 0.88). Figure 1 showed representative maximum intensity projection (MIP) of the whole-body PET images reconstructed with MLAA-based attenuation and CT-based attenuation. Both reconstructions offered similar image quality without any discontinuous artifacts. SUVmean values of 15 patients (presented as average±standard deviation) in different organs were listed in Table 1. Linear correlations between the two methods in different organs were presented in Figure 2. Good correlations between the SUV values per patient in each organ (R2 = 0.93 - 0.99) were observed. Conclusions: MLAA-based reconstruction can estimate activity and attenuation simultaneously without using CT data. PET imaging using MLAA-based reconstruction show comparable image quality and the SUVs show good correlation with that obtained using standard CT-based reconstruction. Key Words: Attenuation correction, MLAA, CT, TOF, 18F-FDGFigure 1. MIP of a patient whole-body 18F-FDG image with MLAA-based reconstruction (left) and CT-based OSEM reconstruction (right).Figure 2. Linear regression analysis between the per patient SUVmean values of MLAA-based reconstruction and CT-based OSEM reconstruction methods in different organs.