Automated Analysis vs. Expert Reading in Nuclear Cardiology: Correlations with the Angiographic Score

Abstract

Background and Objectives: Myocardial perfusion imaging (MPI) has an important role in the non-invasive investigation of coronary artery disease. The interpretation of MPI studies is mainly based on the visual evaluation of the reconstructed images, while automated quantitation methods may add useful data for each patient. However, little evidence is currently available regarding the actual incremental clinical diagnostic performance of automated MPI analysis. In the present study, we aimed to assess the correlation between automated measurements of Summed Stress Score (SSS), Summed Rest Score (SRS) and Summed Difference Score (SDS), with the corresponding expert reading values, using coronary angiography as the gold standard. Materials and Methods: The study was conducted at the Nuclear Medicine Laboratory of the University Hospital of Larissa, Larissa, Greece, οver an one-year period (January 2019–January 2020). 306 patients, with known or suspected coronary artery disease, were enrolled in the study. Each participant underwent a coronary angiography, prior to or after the scintigraphic study (within a three-month period). Either symptom-limited treadmill test, or pharmacologic testing using adenosine or regadenoson, was performed in all participants, and the scintigraphic studies were carried out using technetium 99m (99mTc) tetrofosmin (one-day stress/rest protocol). Coronary angiographies were scored according to a 4-point scoring system (angiographic score; O: normal study, 1: one-vessel disease, 2: two-vessel disease, 3: three-vessel disease). Moreover, automated measurements of SSS, SRS and SDS were derived by three widely available software packages (Emory Cardiac Toolbox, Myovation, Quantitative Perfusion SPECT). Results: Interclass Correlation Coefficients of SSS, SRS and SDS between expert reading and software packages were moderate to excellent. Visually defined SSS, SRS and SDS were significantly correlated with the corresponding results of all software packages. However, visually defined SSS, SRS and SDS were more strongly correlated with the angiographic score, indicating a better performance of expert reading when compared to automated analysis. Conclusions: Based on our results, visual evaluation continues to have a crucial role for the interpretation of MPI images. Software packages can provide automated measurements of several parameters, particularly contributing to the investigation of cases with ambiguous scintigraphic findings.