Alternative statistical models for the examination of clinical positron emission tomography/fluorodeoxyglucose data.

This article describes a method for partitioning metabolic variability found in positron emission tomography/[18F]fluorodeoxyglucose studies. For the 15 subjects examined, 74.8% of the total metabolic variability could be ascribed to individual differences in global metabolic rate, whereas 15.8% of the total variability was consistent regional variation or pattern across subjects. Subsequently, the method of Q-component analysis is described for the identification of strong- and weak-pattern subjects. In addition, a standardization procedure that amplifies the observed pattern by removing systematic individual differences is described. Finally, the implications of these findings and methods for future and clinical studies are discussed.