Background: Quantification of myocardial perfusion single photon emission computed tomography (SPECT) may improve scintigraphic analysis. Recently, a fully operator independent technique for the quantification of myocardial perfusion SPECT was described, based on a normal three-dimensional averaged reference heart. The purpose of this study was to compare the automated SPECT quantification technique with experienced observers.
Methods: A total of 43 patients, 36 with one-vessel coronary artery disease (CAD) and seven with a low likelihood of CAD, underwent 99Tcm-sestamibi SPECT (99Tcm-MIBI SPECT). Three experienced observers and a panel (composed of the three observers), blinded to the clinical and angiographic data, analysed the size and severity of perfusion defects and the relation to the distribution areas of the coronary arteries. Inter-observer agreement was calculated by using kappa (kappa) statistics.
Results: The inter-observer agreement between the human observers and the automated quantitative analysis, for severity and size of perfusion abnormality, was moderate (kappa range 0.38-0.68), while this was fair between three individual observers (kappa range 0.36-0.87) and good between the individual observers and the panel (kappa range 0.63-0.89). There were no differences between the quantitative analysis and the panel in the allocation of perfusion abnormalities to the affected coronary artery.
Conclusions: The operator independent quantification method showed a moderate agreement with individual observers and a panel analysis for size and severity of perfusion abnormalities. The automatic quantification has a similar ability to assign perfusion abnormalities to the diseased coronary artery as compared to an expert panel.