SIMIND scatter estimation: Experimental verification

Scatter estimation and compensation are not new concepts. However a robust and accurate scatter estimation method that easily adapts to all clinically relevant imaging protocols and emission energies are not yet readily available. The goal of this study was to further develop and evaluate the SIMIND scatter estimation (SSE) method. We acquired and simulated two line-sources approximately 8.4 cm apart in a Deluxe Jaszczak phantom as well as acquired the Jaszczak anthropomorphic phantom with activity in the Iowa heart insert and the liver. The measured and simulated data were reconstructed using the rescaled-block iterative (RBI) method using 5 iterations and 4 projections per subset. A three dimensional (3D) Gaussian filter were applied to all data. To compare the different scatter estimation methods, line profiles were drawn through both the reconstructed line-sources and the reoriented short-axis slices of the Iowa heart. The scatter estimates of the methods were also obtained and reconstructed. Profiles for comparison were also drawn. We consistently show that ESSE and TEW underestimate the scatter component compared to SSE, although compensated images look quite similar. However, we agree with others that rigorous modeling all the important aspects of the acquisition process inevitably results in a more accurate and robust representation, especially as we move towards achieving the ultimate goal: absolute quantification.