Body-contour versus circular orbit acquisition in cardiac SPECT: Assessment of defect detectability with channelized Hotelling observer

Abstract

The resolution of a gamma camera is depth-dependent and worsens with increasing distance to the camera resulting in a loss of fine details in SPECT images. A common approach to reduce the effects of this resolution loss is to utilize body-contour acquisition orbits. Even though body-contour orbits can improve resolution of reconstructed images their effect on lesion detection is not well known. To investigate whether body-contour orbits offer better defect detection performance than circular orbits in cardiac SPECT. The mathematical cardiac torso (MCAT) phantom was used to model Tc-sestamibi uptake. A total of four phantoms (two male and two female) with eight defects (four locations and two sizes) were generated and projection data were simulated using an analytical projector with attenuation, scatter, collimator response and acquisition orbit modelling. The circular and body-contour projections were reconstructed using the OSEM algorithm with/without collimator response compensation. Defect detection performance was assessed by calculating area under the receiver operating characteristic (ROC) curve for channelized Hotelling observer. The defect detection performance of circular and body-contour acquisition was very similar and the difference in the area under the ROC curve between the orbits was not statistically significant with or without collimator response compensation. The collimator response compensation, on the other hand, was noticed to be valuable and it provided significantly better defect detection performance than reconstruction without it regardless of the acquisition orbit type. We conclude that by replacing circular orbit with more complex body-contour orbit will not lead to statistically significant increase in defect detection performance in cardiac SPECT.