Effect of varying the scatter window size on the accuracy of TEW scatter correction for Se-153

Objectives Triple-Energy Window (TEW) technique is conventionally used to estimate the scatter under the 103 keV emission of Sm-153 via 93-113 keV photopeak (PP) window, adjacent lower scatter (LS) and upper scatter (US) windows of equal widths. The objective is to evaluate the effect of varying the scatter window size on the accuracy of TEW scatter correction.

Methods Geant4 Monte Carlo was used to simulate the 153Sm energy spectra using a 3/8” NaI-crystal gamma camera with low-energy collimation for three different geometries: hot sphere in cold cylinder (HS-CB), cold sphere in hot cylinder (CS-HB), and hot sphere in hot cylinder with an intensity ratio of 5:1 (HS-HB). The sphere (D=3.3 cm) was located at the center of the cylinder (D=21.04 cm, H=19.24 cm) and 35.26 cm from the surface of the scintillator. The simulated energy spectra were decomposed into different components based on the type of interaction in the source and detector. The PP window was kept constant while the LS (ending at 93 keV) and US (starting at 113 keV) window sizes were varied equally from 3 to 21 keV. The counts in LS and US windows were used to estimate the scatter counts in the PP window via the TEW algorithm. The true scatter counts in the PP window was determined from simulations and used to evaluate the accuracy of the TEW scatter estimate.

Results Non-primary events accounted for 65, 48, and 51 % of the PP window counts for CS-HB, HS-HB, and HS-CB, respectively. The accuracy of TEW scatter estimate showed a strong dependence on the width of the LS and US scatter windows. In all cases, the accuracy of the TEW scatter estimate was highest (3.5, 2.71 and 0.51 % for CS-HB, HS-HB, and HS-CB) for LS and US energy window widths of 10-to-14 keV. The accuracy degraded monotonically when smaller or larger scatter window widths were used. The accuracy of scatter correction was estimated to be 8-10% in the conventional case of a 20 keV energy window.

Conclusions LS and US windows widths of 10-to-14 keV was determined to be most accurate for TEW scatter correction of 153Sm SPECT that yielded a scatter correction accuracy of better than 4%