Skip to Main Content
Cardiac SPECT images are degraded by several physical effects, including attenuation, Compton scatter and collimator geometric response. We proposed a method, analytical SPECT modeling (ASM), for efficiently and accurately modeling the spatially varying SPECT point response function in nonuniform attenuating objects. A quasi Monte Carlo integration method was employed in the model, which resulted in an 1 to 2 orders of magnitude speed up in modeling process. Monte Carlo studies were done on simulated heart images from the NCAT phantom, and our model was implemented in a 3-D OSEM algorithm. The system transition matrix was generated based on the patient-specific attenuation map before reconstruction, which included several parts: 1) the primary matrix which modeled depth dependent collimator response of primary photons; 2) the attenuation factor matrix which modeled nonuniform attenuation of primary photons; 3) the scattering matrix which modeled attenuation, scatter and collimator response of scattered photons. An unmatched projector/backprojector was used to accelerate the reconstruction. Reconstruction results in which different degrading effects were modeled were compared in terms of global quantitative accuracy, contrast and noise level. Results showed that the more complete the degrading effects were modeled in reconstruction, the higher quality and quantitative accuracy could be obtained.