Scatter correction improves concordance in SPECT MPI with a dedicated cardiac SPECT solid-state camera.

Correction for photon attenuation and scatter improves image quality with conventional NaI-based gamma cameras but evaluation of these corrections for novel solid-state dedicated cardiac cameras is limited. In this study, we assess the accuracy of dual-energy-window (DEW) scatter correction (SC) applied to clinically acquired (99m)Tc-tetrofosmin myocardial perfusion images obtained on a dedicated multi-pinhole camera with cadmium-zinc-telluride (CZT) detectors (GE Discovery NM530) compared to DEW scatter-corrected images from our conventional SPECT camera (GE Infinia Hawkeye 4; INF). A modified DEW SC method was formulated to account for the detection of primary photons in the lower energy window (120 keV ± 5%) with CZT detectors, in addition to estimating the scattered photons detected in the photopeak window (140 keV ± 10%). Phantom experiments were used to estimate the DEW correction parameters. Data from 108 patients, acquired using a standard rest/stress Tc-99m-tetrofosmin SPECT/CT protocol on both cameras, were reconstructed with no correction (NC), attenuation correction (AC), and AC with DEW-SC. Images were compared based on the summed stress/rest/difference scores (SSS/SRS/SDS) calculated by clinical software. The correlation between SSS/SRS for the two cameras was excellent (r ≥ 0.94). The mean difference between cameras was <0.4 for SSS/SRS/SDS scores. Since datasets did not follow a normal distribution, non-parametric tests were used to show significant differences between datasets. Classification of disease (SSS) was highly correlated, as ranked by the two cameras (kendall's tau = 0.72, P < .001). AC significantly reduced the mean difference between the two cameras for SSS/SRS compared to NC. AC without SC on the CZT introduced a bias towards higher scores when compared to the INF, which was reduced after applying SC. Although SC increased noise, the scores for the AC/SC images were not significantly different between the two cameras (P > .1). DEW-SC on the CZT camera was feasible and produced images that are not significantly different from those acquired on the INF camera. Although use of SC on CZT images does increase noise, the resultant noise does not introduce bias relative to the INF camera.