PET system calibration and attenuation correction

A statistical model involving machine calibration factors in PET is developed based on the analysis of the correlation structure of the blank scan data. These factors include the nonuniform detector efficiencies and the geometric effects. The crystal interference effect is shown to be the main feature of the correlation function. A likelihood based theory is developed and the maximum likelihood estimates of the factors are derived. In the application of the model, estimated machine effects are used to calibrate the transmission scan and a robust smoothing is then applied to get the attenuation correction factor. The proposed technique is illustrated with studies on the GE Advance PET Scanner.