A new method for modeling the spatially-variant, object-dependent scatter response function in SPECT

Scatter compensation using iterative reconstruction results in improved image quality and quantitative accuracy compared to subtraction-based methods, However, this requires knowledge of the spatially-varying, object-dependent scatter response function (SRF), We have previously developed a method, slab derived scatter estimation (SDSE) for estimating the SRF. However, this method has reduced accuracy for nonuniform attenuators and Tl-201 imaging. In this paper we present a new method which provides better modeling of the SRF for Tl-201 SPECT, and should provide improved accuracy for nonuniform attenuators. The method requires 3 image space convolutions and an attenuated projection for each viewing angle. Implementation in a projector-backprojector pair for use with an iterative reconstruction algorithm would require 2 image space Fourier transforms and 6 image space inverse Fourier transforms per iteration. We observed good agreement between SRFs and projection data estimated using this new model compared to those obtained using Monte Carlo simulations