Comparison among Reconstruction Algorithms for Quantitative Analysis of 11C-Acetate Cardiac PET Imaging.

Ximin Shi,Haiyan Ding,Yonghong Dang,Li Huo,Shuai Liu,Fang Li,Nan Li,Jie Cui,Hui Zhang,Guilan Hu,Yue Zhang

doi:10.1155/2018/9193403

Abstract

Objective Kinetic modeling of dynamic 11C-acetate PET imaging provides quantitative information for myocardium assessment. The quality and quantitation of PET images are known to be dependent on PET reconstruction methods. This study aims to investigate the impacts of reconstruction algorithms on the quantitative analysis of dynamic 11C-acetate cardiac PET imaging. Methods Suspected alcoholic cardiomyopathy patients (N = 24) underwent 11C-acetate dynamic PET imaging after low dose CT scan. PET images were reconstructed using four algorithms: filtered backprojection (FBP), ordered subsets expectation maximization (OSEM), OSEM with time-of-flight (TOF), and OSEM with both time-of-flight and point-spread-function (TPSF). Standardized uptake values (SUVs) at different time points were compared among images reconstructed using the four algorithms. Time-activity curves (TACs) in myocardium and blood pools of ventricles were generated from the dynamic image series. Kinetic parameters K1 and k2 were derived using a 1-tissue-compartment model for kinetic modeling of cardiac flow from 11C-acetate PET images. Results Significant image quality improvement was found in the images reconstructed using iterative OSEM-type algorithms (OSME, TOF, and TPSF) compared with FBP. However, no statistical differences in SUVs were observed among the four reconstruction methods at the selected time points. Kinetic parameters K1 and k2 also exhibited no statistical difference among the four reconstruction algorithms in terms of mean value and standard deviation. However, for the correlation analysis, OSEM reconstruction presented relatively higher residual in correlation with FBP reconstruction compared with TOF and TPSF reconstruction, and TOF and TPSF reconstruction were highly correlated with each other. Conclusion All the tested reconstruction algorithms performed similarly for quantitative analysis of 11C-acetate cardiac PET imaging. TOF and TPSF yielded highly consistent kinetic parameter results with superior image quality compared with FBP. OSEM was relatively less reliable. Both TOF and TPSF were recommended for cardiac 11C-acetate kinetic analysis.

Highlights

Dynamic PET imaging with kinetic analysis provides more quantitative information compared with the commonly used standardized uptake value (SUV) [1, 2]
In order to minimize the difference in coefficient of variance (CV) and obtain similar image qualities, a 3D Gaussian filter of 5.0 mm in FWHM was applied to both TOF and time-of-flight and point-spread-function (TPSF) images, while a 3D Gaussian filter of 6.5 mm in FWHM was applied to ordered subsets expectation maximization (OSEM) images
Kinetic parameters K1 and k2 exhibited no statistical difference among the four reconstruction algorithms in terms of mean value and standard deviation

Summary

Introduction

Dynamic PET imaging with kinetic analysis provides more quantitative information compared with the commonly used standardized uptake value (SUV) [1, 2]. Utilizing more accurate physical system model, OSEM can greatly reduce the noise in the reconstruction and significantly improve the image quality [8, 9]. Progress has been made on OSEM-type reconstruction by combining with time-of-flight (TOF) information [10, 11] and point-spread-function (PSF) kernel [12, 13] in the iterative reconstruction process. These new techniques substantially improve image quality and reduce the partial volume effect (PVE) [14]

Methods

Results

Discussion

Conclusion