Possibilities of software phantoms for quality control of KBS in nuclear medicine.

The assessment of the results of a "knowledge-based system" (KBS) for quality control is a basic requirement for clinical application: Large numbers of test studies are necessary in order to cover as widely as possible the spectrum of cases to be analyzed by the KBS. The use of original patient data as test data is one possibility, but real data are provided unevenly. This is due to the set of characteristics which are relevant to the analysis. Data are available in a limited quantity only. This implies a remaining set of unvalidated cases which are not represented in the data pool. The software phantom is an approach towards systematically guided validation. It permits the generation of test data adjusted to the demands of the validation.