Scatter restoration in PET imaging

Positron emission tomography (PET) is a quantitative tool having the capability of estimating physiological parameters in vivo. However, in order for these parameters to be accurate, PET data need to be corrected for image degrading effects such as scatter. The amount of scatter and its axial and transaxial distributions in the images depend on the position of the emitting sites, on the scattering object and on the collimators. Generally scatter functions are determined from point sources, or by extrapolation or the radioactivity distribution from out of the object in the image, or by analytical estimation of single scatter based on emission, transmission and photon detection. In this work, scatter fraction is determined by Monte Carlo calculations based on PET images in humans and in rats measured with the Philips Allegro scanner and the Sherbrooke small animal scanner, respectively. Assuming the image slice is made of tissue only, the scatter fraction estimated as a function of the number of Compton interactions in human (rat): no scatter: 9.75% (66.16%), single scatter: 21.92% (26.96%), double scatter (both photons scatter once): 12.42% (2.71%), multiple scatter: 55.89% (4.18%). Moreover, the distribution of each of these types of scatter has its characteristics, generally having its maximum corresponding to source location. In conclusion, scatter functions need to be determined as a function of the type of scatter in order to process an accurate scatter correction.