Effects of energy space smoothing and projection space normalization on multispectral PET image quality

Independent processing of multispectral positron emission tomography (MSPET) data in individual energy frames has the potential to improve system sensitivity and the accuracy of energy dependent scatter correction. However, statistical fluctuations due to the use of multiple energy windows and low system detection efficiency severely undermine this potential. These limitations have been overcome without resolution loss by smoothing data in the energy space to suppress statistical fluctuations and by normalizing detector efficiency in the spatial domain to minimize systematic errors. The effectiveness of these corrections was evaluated by comparing images acquired in different energy frames with and without energy space smoothing. Smoothing improved the sharpness and contrast and decreased noise of images. The FWHM and FWTM evaluated from line source images confirmed an earlier postulate which stated that smoothing in the energy space has no effect on image resolution since such process does not move counts across lines of response (LORs). It was concluded that smoothing in the energy space in conjunction with normalization in the projection space is a prerequisite for subsequent energy-dependent data processing such as scatter correction.