Estimation of Scatter Function in SPECT Imaging of Heart Using Deconvolution Method: a simulation study

Scattered photons are one of the main causes of degrading the contrast of lesions and resolution in SPECT imaging of the heart that result in error in quantification. The usual technique for the rejection of scattered photons is through energy windowing. However, because of limited energy resolution of current scintillation cameras it is impossible to avoid scatter photons from detection. Modeling of Compton scattering through finding suitable functions was proposed in this study. These functions were used for scatter correction through deconvolution in next step. Monte Carlo simulation was used for creating projections of three different activity sources: a line source passed through left ventricle, a point source placed in left ventricle and finally real activity distribution of Tc-99m in torso organs. All of these sources were placed in a digital attenuation phantom which modeled a real patient body. Images of primary and scattered photons were acquired separately. Convolution and 2D deconvolution in Fourier domain was applied for estimating the primary projections through total ones. In the first step, scatter and total images were modeled as convolution of a modified exponential function with primary image. The best exponent value was determined for each of 64 views (0.115 to 0.150 according to heart to detector distance). In the next step, these functions were used for scatter correction through deconvolution. Sum of the square differences between the primary and scatter corrected images were decreased considerably, myocardium to cavity contrast increased for all of 64 views (34% ± 10%). Good agreement between the real primary and scatter corrected images were also found.Discussion and These results indicate that deconvolution technique for the scatter compensation can significantly reduce the degrading roles of scattering in quantitative SPECT imaging.