Effects of statistical noise on graphic analysis of PET neuroreceptor studies

Because of its computational simplicity, the graphic method introduced by L ogan et al. is frequently used to analyze time-activity curves of reversibl e radiotracers measured in brain regions with PET. The graphic method uses a nonlinear transformation of data to variables that have an asymptotically linear relationship. Compared with compartmental analysis of untransformed data, the graphic method enables derivation of regional distribution volum es that are free from assumptions about the underlying compartmental config uration. In this article, we describe statistical bias associated with this nonlinear transformation method. Methods: Theoretic analysis, Monte Carlo simulation, and statistical analysis of PET data were used to test the grap hic method for bias. Results: Mean zero noise is associated with underestim ation of distribution volumes when data are analyzed with graphic analysis, whereas this effect does not occur when the same data are analyzed by nonl inear regression and compartmental analysis. Moreover, this effect depends on the magnitude of the distribution volume, so that the bias is more prono unced in regions with high receptor density than regions with low receptor density or no receptors (region of reference). Conclusion: These results in dicate that conventional kinetic analysis of untransformed data is less sen sitive to mean zero noise than is graphic analysis of nonlinearly transform ed data.