Three-dimensional SPECT reconstruction with transmission-dependent scatter correction

Abstract

The quality of single-photon emission computed tomography (SPECT) imaging is hampered by attenuation, collimator blurring, and scatter. Correction for all of these three factors is required for accurate reconstruction, but unfortunately, reconstruction-based compensation often leads to clinically unacceptable long reconstruction times. Especially, efficient scatter correction has proved to be difficult to achieve. The objective of this article was to extend the well-known transmission-dependent convolution subtraction (TDCS) scatter-correction approach into a rapid reconstruction-based scatter-compensation method and to include it into a fast 3D reconstruction algorithm with attenuation and collimator-blurring corrections. Ordered subsets expectation maximization algorithm with attenuation, collimator blurring, and accelerated transmission-dependent scatter compensation were implemented. The new reconstruction method was compared with TDCS-based scatter correction and with one other transmission-dependent scatter-correction method using Monte Carlo simulated projection data of (99m)Tc-ECD and (123)I-FP-CIT brain studies. The new reconstruction-based scatter compensation outperformed the other two scatter-correction methods in terms of quantitative accuracy and contrast measured with normalized mean-squared error, gray-to-white matter and striatum-to-background ratios, and also in visual quality. Highest accuracy was achieved when all the corrections (i.e., attenuation, collimator blurring, and scatter) were applied. The developed 3D reconstruction algorithm with transmission-dependent scatter compensation is a promising alternative to accurate and efficient SPECT reconstruction.