Evaluation of Triple-energy Window Method with Two Approximations for Scatter Correction Studies in Single-photon...

Simulations using Monte Carlo packages are increasingly used in nuclear imaging for positron emission tomographic and single-photon emission computed tomographic applications, either for modeling imaging systems or developing algorithms for analysis improvement of image quantifications. This work implemented a GEANT4-based single-photon emission computed tomographic simulation system, GATE, to evaluate the performance of two scatter correction methods: the triple-energy window method using trapezoidal approximation and the triple-energy window method using triangular approximation. A cold-spot hot-background phantom and a resolution phantom were simulated to generate projection images for scatter correction. Both methods were not found to have strictly valid assumptions. A comparative assessment of these two methods was made based on image contrast improvement, image mottle level, and image resolution improvement. Results showed that triangular approximation was better than trapezoidal approximation for contrast improvement with greater image mottle level and some improvement in the resolution of the resolution phantom for both approximations, but trapezoidal approximation with these criteria was better than triangular approximation. A compromise between the contrast improvement and the image mottle level exists. Neither of the two methods performed best for all three criteria.