Background: The increased use of hybrid SPECT-CT systems in myocardial perfusion imaging calls for a careful review of protocols that are employed in data acquisition and processing. Our study investigates the cases of potential false perfusion defects that may appear in cardiac images reconstructed with CT-based attenuation correction and high-resolution acquisition matrix.
Methods: Phantom experiments performed on Infinia-Hawkeye (GE Healthcare) modeling patients with normal hearts and different body sizes and shapes were reconstructed using standard ordered subsets expectation maximization (OSEM) method with 10 subsets and 2-50 iterations. The CT-based attenuation correction (AC) with and without distance-dependent resolution recovery (RR) were employed.
Results: Images reconstructed from scans with a standard thorax phantom did not show any artifacts. However, some images reconstructed from the data with extra water bags (modeling high and non-uniformly distributed attenuation) clearly displayed false perfusion defects. A potential for creating such artifacts was especially high in complex reconstruction cases. In most cases, the severity of defects decreased when reconstructions with more iterations were used.
Conclusions: In situations with strong and non-uniform attenuation, the reconstruction methods with sophisticated data processing (large matrix size, AC + RR corrections) may require substantially more processing than is currently used and/or recommended.