Correction for scatter using a dual energy window technique with a tomograph operated without septa

A method of scatter correction using dual energy window acquisition was studied on a Siemens/CTI 953B brain scanner operated without septa. The basis of the method is to determine the scatter component by direct measurement using improved block detector spectroscopy. Unscattered events are derived from a knowledge of the ratios, between the two windows, of scattered events (R/sub sc/) and unscattered events (R/sub unsc/) The upper (photopeak) window was set at 380-850 keV and the lower window accepted coincidences where one or both photons deposited energy between 200 and 380 keV. The choice of windows is a compromise between maximizing counting statistics and minimizing the spatial variation of R/sub unsc/. Transaxially R/sub sc/ does not vary significantly over the central 10 cm of a 20-cm cylinder either for a central line source or uniformly distributed activity. The variation is similar for a 15-cm cylinder with the absolute level about 10% lower. R/sub unsc/ is practically flat over the 40-cm transaxial FOV (field of view). There is a systematic axial variation in R and R/sub unsc/ which follows the block detector structure but which can be allowed for by interplane normalization. In phantoms with asymmetric and uniform activity distributions this method of correction shows good improvement in contrast.<>