Performance evaluation of a small field-of-view camera for myocardial perfusion SPECT using a channelized Hotelling observer

Abstract

This study compares the performance of cameras with small (20/spl times/20 cm) and standard (50/spl times/40 cm) fields-of-view for myocardial perfusion SPECT. Simulated data from four MCAT phantom anatomies (two male and two female) and 6 defect locations were used. Projection data were generated using an analytic projector that includes the effects of attenuation, scatter, and collimator-detector response. Poisson noise was simulated. For the small camera, the center-of-rotation (COR) was positioned both at the center of the heart and at 13 shifted positions resulting id different degrees of truncation. The noisy projection data were reconstructed using the filtered backprojection and iterative OS-FM algorithms with no compensation for image degrading effects. The channelized Hotelling Observer (CHO) was applied centered to the myocardial defect in the short-axis images. The rating data generated by the CHO were analyzed using receiver operating characteristics (ROC) methodology. The area under the ROC curve was used to compare the two cameras and the two reconstruction algorithms. If the heart is not truncated, the two cameras perform similarly with respect to the task of detecting myocardial perfusion defects. In cases where truncation is near the heart, the standard camera has better performance, and OS-FM improves defect detectability.