ABSTRACT
The determination of time-activity curves based on measurements of the biokinetics at a few time points is a common task in Nuclear Medicine. In this chapter, the input, the processing and the output are presented and discussed. The first inputs are the measured data, but also data representing a priori knowledge about the sought biokinetics. The second input consists of the mathematical models used: These include not only different types of pharmacokinetic models, but also the models for the uncertainty in the measured data. Both inputs are used for the curve-fitting process. For this fitting process, the definition of an adequate objective function and (user-independent) quality criteria for the evaluation of the fits are discussed. As in most cases the true mathematical model to be fitted to the data is unknown, the model selection procedure is presented for the case of employing the Akaike Information Criterion. From this, we finally get the 3 outputs we are looking for, i.e. the selected model that is most supported by the data, the fitted parameter values of the model and – not to forget – their uncertainty.
