Scatter correction in scintillation camera imaging of positron-emitting radionuclides

The use of Anger scintillation cameras for positron SPECT has become of interest recently due to their use with imaging 2-/sup 18/F deoxyglucose. Due to the special crystal design (thin and wide), a significant amount of primary events will also be recorded in the Compton region of the energy spectra. Events recorded in a second Compton window (CW) can add information to the data in the photopeak window (PW), since some events are correctly positioned in the CW. However, a significant amount of the scatter is also included in CW which needs to be corrected. This work describes a method whereby a third scatter window (SW) is used to estimate the scatter distribution in the CW and the PW. The accuracy of estimation has been evaluated by Monte Carlo simulations in a homogeneous elliptical phantom for point and extended sources. Two examples of clinical application are also provided. Results from simulations show that essentially only scatter from the phantom is recorded between the 511 keV PW and 340 keV CW. Scatter projection data with a constant multiplier can estimate the scatter in the CW and PW, although the scatter distribution in SW corresponds better to the scatter distribution in the CW. The multiplier k for the CW varies significantly more with depth than it does for the PW. Clinical studies show an improvement in image quality when using scatter corrected combined PW and CW.