SPECT scatter correction algorithms using position and energy information: preliminary simulation studies

Summary form only given, as follows. Numerous authors have presented methods for scatter correction in SPECT (single photon emission computed tomography) that make partial use of energy information. The present authors are currently testing a board that saves energy and position for each detected event, and have begun evaluating (using simulation) algorithms for scatter reduction that could make use of such information. Their current approach is to collect an energy spectrum at each point in the image. The pointwise spectra are then deconvolved by the energy PSF of the detection system to estimate the primary and scatter components at each point. Parametric modeling of the scatter spectrum makes this step quite fast. The scatter components are spatially smoothed in an energy-dependent fashion, and the smoothed scatter estimates are subtracted from the raw data, which are reconstructed using conventional methods. Simulated images and findings relating to optimal smoothing and deconvolution methods have been obtained. Some cases of the algorithms considered can be implemented as simple weighted energy acquisition methods.<>