Monte Carlo evaluation of accuracy and noise properties of two scatter correction methods

Two independent scatter correction techniques, transmission dependent convolution subtraction (TDCS) and the triple-energy window (TEW) method, were evaluated in terms of quantitative accuracy and noise properties using Monte Carlo simulation (EGS4). Emission projections (primary, scatter and scatter plus primary) were simulated for ^99m Te and ²⁰¹Tl for numerical chest phantoms. Data were reconstructed with an ordered-subset ML-EM algorithm including attenuation correction using the transmission data. In the chest phantom simulation, TDCS provided better S/N than TEW, and better accuracy, i.e., 1.0% vs. -7.2% in myocardium, and -3.7% vs. -30.1% in the ventricular chamber for ^99mTc with TDCS and TEW, respectively. For ²⁰¹Tl, TDCS provided good visual and quantitative agreement with a simulated true primary image without noticeably increasing the noise after scatter correction. Overall the TDCS proved to be more accurate and less noisy than TEW, facilitating quantitative assessment of physiological functions with SPECT