Statistical model-based scatter correction for 3D PET

The aim of this study is to propose a scatter correction method based on a statistical model of mixture Gaussian Model (MGM) that has robust capability for information extraction and can make smaller in size of PET data stored. Clustering algorithm based on MGM is applied to extract scatter component. MGM assumed that measured data are sum of unknown components that statistical characteristic is Gaussian. MGM was applied to three different implementations of one-, two-, and three-dimensional projection data. Good extraction performance and reduction of the amount of data are archived.