摘要目的 探讨多针孔碲锌镉(CZT)心脏专用SPECT仪(CZT-SPECT)行心肌动态SPECT显像时无完整物理校正对心肌血流绝对定量结果的影响.方法 收集阜外医院2018年7月至2019年1月间30例[男18例,女12例;年龄(63±9)岁]疑似或已知冠状动脉粥样硬化性心脏病的患者行心肌动态SPECT显像.采用不同物理校正模式处理图像:无校正、部分校正[噪声消减校正、(噪声消减+散射)校正、(噪声消减+散射+空间分辨率恢复)校正]及完整校正[(噪声消减+散射+空间分辨率恢复+组织衰减)校正].动力学建模采用单组织双腔室模型,通过数据-模型拟合度(R2)及血池溢出分数(FBV)进行质量评价;比较无校正、部分校正与完整校正的静息心肌血流量(RMBF)、负荷心肌血流量(SMBF)及心肌血流储备(MFR)的差异.采用Wilcoxon符号秩检验和直线回归分析行统计学处理.结果 在所有数据集中,与完整校正相比,无校正结果的一致性最低(R2:静息=0.69,负荷=0.78;z值:4.78和4.78,均P<0.01),左心室血池溢出效应最高(FBV:静息=0.37,负荷=0.40;z值:-3.40和-3.30,均P<0.01).增加校正项后,不同校正状态的结果与完整校正结果的一致性得到改善,FBV逐步降低.与完整校正相比,无校正高估了SMBF和MFR(z值:1.27和-3.50,均P<0.01),部分校正均高估了RMBF和SMBF(z值:-4.55～1.27,均P<0.01),噪声消减及(噪声消减+散射)校正均低估了MFR(均P<0.05).线性回归分析显示,无校正、噪声消减校正、(噪声消减+散射)校正、(噪声消减+散射+空间分辨率恢复)校正与完整校正RMBF的回归系数为0.908～1.210,Bland-Altman分析显示正向或逆向的偏差(偏差分别为-0.07、0.21、0.26和0.15 ml·min-1·g-1);上述校正与完整校正SMBF的回归系数为1.129～1.308,Bland-Altman分析显示正向偏差(偏差分别为0.60、0.25、0.28和0.24 ml·min-1·g-1);上述校正与完整校正MFR的回归系数为0.907～1.318,Bland-Altman分析显示极小正向或逆向的偏差(偏差分别为0.70、-0.11、-0.05和0.01).结论 采用多针孔CZT-SPECT行心肌血流定量时,在动态SPECT图像后处理过程中,完整的物理校正能提高数据-模型的一致性,降低左心室血池溢出效应,进而提高心肌血流绝对定量的准确性.

abstractsObjective To investigate the impact on myocardial blood flow (MBF) quantitation with multi-pinhole cadmium zinc telluride (CZT)-SPECT with or without partial physical corrections. Methods A total of 30 patients (18 males, 12 females; age: (63±9) years) with suspected or known coronary heart diseases who underwent dynamic SPECT from July 2018 to January 2019 in Fuwai Hospital were enrolled. Images were reconstructed using different corrections: no correction (NC), partial corrections ((noise re-duction ( NR) , NR+scatter correction ( SC) , NR+SC+resolution recovery ( RR) ) , NR+SC+RR+attenua-tion correction ( AC;total corrections, TC) . Kinetic modeling integrated one-tissue two-compartment model while using index of fitting quality ( R2 ) and fraction blood volume ( FBV) to assess the quality of modeling. Rest MBF ( RMBF) , stress MBF ( SMBF) and myocardial flow reserve ( MFR) quantified from no correc-tion ( NC) or partial corrections were compared with those of TC. Wilcoxon signed rank test and linear re-gression analysis were used to analyze the data. Results Compared to TC, NC showed the lowest R2( rest:0.69, stress:0.78;z values:4.78 and 4.78, both P<0.01) and highest FBV ( rest:0.37, stress:0.40;z values: -3.40 and -3.30, both P<0.01). The improvement of R2 and FBV was consistent with increased corrective terms. Compared with TC, NC overestimated SMBF and MFR ( z values:1.27 and-3.50, both P<0.01), all partial corrections overestimated RMBF and SBMF (z values:from -4.55 to 1.27, all P<0.01). NR and NR+SC underestimated MFR (both P<0.05). Linear regression analysis showed that the regressive coefficients of RMBF between NC, NR, NR+SC, NR+SC+RR and TC were 0.908-1.210, and Bland-Altman plots of RMBF demonstrated positive or negative biases (-0.07, 0.21, 0.26, 0.15 ml·min-1·g-1). The regression coefficients of SMBF were 1. 129-1. 308, and Bland-Altman plots demonstrated positive biases (0. 60, 0.25, 0.28, 0.24 ml·min-1·g-1). The regression coefficients of MFR were 0.907-1.318, and Bland-Altman plots demonstrated positive or negative biases (0.70,-0.11,-0.05, 0.01). Conclusion Full physical corrections can improve the index of fitting quality in the kinetic modeling and reduce left ventricle spillover, which help to warrant the accuracy of SPECT myocardial blood flow quantitation with multi-pin-hole CZR-SPECT.