Quantitative accuracy of SPECT imaging with a dedicated cardiac camera: Physical phantom experiments

Abstract

Recently, there has been increased interest in dedicated cardiac SPECT scanners with multi-pinhole designs and improved detector technology. However, the pinhole collimator design introduces position-dependent attenuation, sensitivity and resolution variations. Variations in attenuation patterns and energy-spectrum responses can also inhibit accurate measurement of activity distributions. Simple correction methods for these effects are easily implemented however their level of accuracy is unclear. In this study, we assess the quantitative accuracy and reproducibility of absolute activity measurements made using easily implemented correction techniques applied to controlled physical phantom experiments. Methods. Activity inserted into the cardiac compartment of an Anthropomorphic Torso phantom was measured through multiple 99mTc-sestamibi acquisitions (n=10). Experiments were repeated with- (hot) and without activity in the torso compartment (cold). A dose calibrator was used to measure the true net activity inserted into each compartment. A ten minute SPECT image was acquired using a dedicated multipinhole cardiac camera with CZT detectors (GE Discovery NM530c), followed by a CT scan for attenuation correction (AC). Acquired data were reconstructed with no corrections (NC), AC and AC with dual energy window (DEW) scatter correction (ACSC). Results. T-tests comparing hot- and cold torso datasets showed no significant mean difference for all three reconstruction methods (AC, NC, ACSC: hot vs cold-torso) (p>0.07). AC and ACSC significantly reduced the mean error over NC (p<0.001). The mean AC and ACSC errors were significantly different for the hot-torso (p=0.04) but not for the cold-torso (p=0.09). While DEW-SC seemed to increase measurement uncertainty compared to AC in both cases, the increase was not significant (p ≥0.2). Both AC and ACSC significantly increased uncertainty over NC for the cold-torso-phantom (p<0.03), but not for the hot-torso case (p>0.1). Conclusions. CT-AC and an easily implemented DEW-SC significantly improve quantitative measurement of 99mTc-SPECT myocardial activity with a dedicated cardiac camera (error <=2.2+/-6.5%).