图像重建方法对门控核素心肌灌注显像室壁相位分析的影响

Effect of different image reconstruction algorithms on phase analysis of gated myocardial perfusion SPECT studies

目的 评价几种常用的SPECT门控核素心肌灌注显像（GSPECT MPI）图像迭代重建方法对相位分析的影响。方法对所选的30例患者应用Philips Cardio MD系统采集GSPECTMPI图像，分别使用滤波反投影（FBP）、最大似然一最大期望值（MLEM）、带三维分辨率恢复的MLEM（AST）、带衰减校正（AC）的MLEM及带AC和蒙特卡罗散射校正（ACSC）的MLEM对GSPECT数据进行重建。将重建数据传递至SyncTool，以测量左心室不同步参数（相位标准差和直方图带宽）。使用配对t检验比较由FBP和以上各种迭代法所得到的左心室不同步参数。结果负荷GSPECTMPI相位分析结果中由FBP、MLEM、带AC的MLEM、带ACSC的MLEM、AST所得到的相位标准差分别为11．6°，10．9°，11．2°，11．6°，11．4°；直方图带宽分别为35．7°，34．3°，35．1°，36．9°，35．1°。静息GSPECTMPI相位分析结果中由上述5种方法所得到的相位标准差分别为15．2°，14．5°，15．4°，15．4°，14．8°；直方图带宽分别为47．3°，46．4°，46．4°，47．9°，46．1°。负荷显像时从各迭代方法和FBP所得到的左心室不同步参数之间差异无统计学意义（t值-1．179～1．554，P均〉0．05），静息显像时各参数间差异亦无统计学意义（t值-0．714—0．666，P均〉0．05）。结论标准FBP重建已足够用于精确的相位分析，SyncTool测量左心室不同步的技术可广泛应用于临床。

Objective To evaluate the effect of various iterative reconstruction methods on phase analysis of gated myocardial perfusion imaging （MPI）. Methods Thirty consecutive patients scanned by the Philips CardioMD system were recruited into this study. The gated SPECT （GSPECT） data were reconstructed with filtered backprojection （FBP） , maximum likelihood expectation maximization （MLEM） , three-dimensional （3D） resolution recovery MLEM （AST）, attenuation corrected （AC） MLEM, AC and 3D Monte Carlo scatter corrected （ACSC） MLEM methods. Parameters of left ventricular （LV） dyssynchrony （phase standard deviation and histogram bandwidth） were measured using the software SyncTool. Paired t-test was used to compare the differences of the LV dyssynchrony indices between FBP and MLEM, AC MLEM, ACSC MLEM, AST respectively. Results The phase standard deviations of stress GSPECT MPI for FBP, MLEM, AC MLEM, ACSC MLEM, and AST were 11.6°, 10.9°, 11.2°, 11.6° , 11.4° respectively; while the histogram bandwidths were 35.7°, 34.3°, 35.1°, 36.9°, 35.1° respectively. The phase standard deviations of rest GSPECT MPI for FBP, MLEM, AC MLEM, ACSC MLEM and AST were 15.2°, 14. 5°, 15.4°, 15.4°, 14.8° respectively; while the histogram bandwidths were 47.3°, 46.4°, 46.4°, 47.9°, 46.1° respectively. No statistical significance was observed between the FBP and various iterative reconstruction methods for both the stress and rest GSPECT MPI study （t： - 1. 179 to 1. 554, P 〉 0.05 for all）. Conclusion The standard FBP reconstruction method is accurate enough for the measurement of LV dyssynchrony indices using the widely used clinical software SyncTool.