Compensation for Distance-Dependent Resolution in Cardiac-Perfusion SPECT: Impact on Uniformity of Wall Counts and Wall Thickness

We investigated the impact on the uniformity of the polar map of the maximum wall count, and on the cardiac wall thickness of several methods used to compensate for distancedependent resolution (DDR) in cardiac perfusion imaging. Compared were: 1) filtered backprojection (FBP) reconstruction with no correction for attenuation and DDR; 2) FBP with Bellini attenuation compensation (AC); 3) FBP with Bellini AC and frequency distance relationship (FDR) restoration filtering prior to reconstruction; 4) ordered-subset maximum likelihood estimation-maximization (OSEM) reconstruction with just AC; and 5) including DDR compensation and AC in OSEM reconstruction with 3D postreconstruction Gaussian filtering. The best uniformity was obtained with OSEM reconstruction which included just AC. Nearly the same uniformity could be obtained with reconstruction by OSEM which included correction for both AC and DDR when strong 3D post-reconstruction Gaussian filtering was included. With application of this filtering OSEM with compensation for both AC and DDR achieved significantly better recovery of the original wall thickness at the expense of higher noise magnitude. For the implementations compared herein, it was noted that the modeling of AC and DDR in OSEM yielded a slightly better combination of image quality metrics than FDR restoration. Observer detection studies are needed to determine if compensation for DDR and AC does result in improved lesion detection over just compensation for AC.