Reconstruction-Based Estimation of the Scatter Component in Positron Emission Tomography

A new method to assess the scatter component in positron emission tomography (PET) based on estimating the low-frequency component corresponding to scattered events using ordered subsets - expectation maximization (OSEM) reconstructions is proposed in this paper and evaluated using Monte Carlo simulation studies, experimental phantom measurements and clinical studies. The rationale of this method called Statistical Reconstruction-Based Scatter Correction (SRBSC) is that the image corresponding to scattered events in the projection data consists of almost low-frequency components of activity distribution and that the low-frequency components will converge faster than the high-frequency ones in successive iterations of statistical reconstruction methods such as OSEM. The second assumption is that the high-frequency components will be smeared, i.e. filtered by the scatter response kernels. A simple model has been devised to parameterize the scatter component using Monte Carlo simulations. The unscattered component estimated using SRBSC was compared to the true unscattered component as estimated by Monte Carlo simulations for simple phantom geometries and clinically realistic source distributions. Quantitative analysis was also performed on reconstructed images using simple metrics like the contrast, absolute concentration, recovery coefficient and signal-to-noise ratio. The SRBSC method tends to undercorrect for scatter in most regions of the 3D Hoffman brain phantom, but gives good activity recovery values which average within 1%. It was concluded that the proposed method improves image quality and the contrast compared to the case where no correction is applied and that an accurate modeling of the scatter component is essential for a proper scatter correction.