Comparison of myocardial perfusion and functional data of LV between 3D OSEM and FBP reconstructed images on gated myocardial perfusion SPECT

Recently, 3D iterative reconstruction (IR) of myocardial perfusion imaging (MPI) has attracted attention due to the superior image quality produced. This may increase the accuracy of image interpretation. And the effect of iterative reconstruction on left ventricular ejection (LVEF) fraction, end diastolic (EDV), and end systolic volumes (ESV) has not been evaluated. This study was performed to evaluate the effect of 3D IR of gated myocardial perfusion SPECT (gSPECT) on accuracy of image interpretation and quantitative functional information (QFI) in comparison to the conventional filtered back-projection (FBP). Thirty six patients (pts, 58.810.2 yrs, M: F=24: 12) with wide range of LVEF (30-74%) on previously acquired one day rest-stress Tc-99m tetrofosmin (29 pts) or TI-201 stress-redistribution (7 pts) gSPECT scans were selected. All of them had underwent coronary angiography and echocardiography within 2 months interval. The studies were processed by the standard FBP and the 3D IR ordered-subset expectation-maximization (OSEM) methods, respectively. The MPI interpretation with FBP and IR was confirmed by coronary angiography (= 50% stenosis). The QFI were calculated by using Quantitative Gated SPECT (QGS) with both FBP and IR processing. The sensitivity, specificity, PPV, NPV and accuracy of MPI with FBP on per patient analysis are 86.2%, 85.7%, 96.1%, 60.0% and 86.1% respectively. On the other hand with IR 86.2%, 71.4%, 92.5%, 55.5% and 83.3% respectively. There was no significant difference between FBP and 3D IR on per vessel analysis. Excellent correlation of QFI between the two methods was noted with r = 0.99 for both stress and rest LVEF, EDV, and ESV. In our study 3D IR of MPI showed no significant difference in the accuracy of MPI interpretation. Quantitative functional results were comparable using FBP and IR over a wide range of LVEFs