Evaluation of the ICS and DEW scatter correction methods for low statistical content scans in 3D PET

The performance of the Integral Convolution and the Dual Energy Window scatter correction methods in 3D PET has been evaluated over a wide range of statistical content of acquired data (1 M to 400 M events). The order in which scatter correction and detector normalization should be applied has also been investigated. Phantom and human neuroreceptor studies were used with the following figures of merit: axial and radial uniformity, sinogram and image noise, contrast accuracy and contrast accuracy uniformity. Both scatter correction methods perform reliably in the range of number of events examined. Normalization applied after scatter correction yields better radial uniformity and fewer image artefacts.