An intercomparison of the accuracy of computer planning systems for brachytherapy.

The accuracy of brachytherapy computer planning systems has been investigated by means of an intercomparison of the dose distributions as calculated by these systems for five specific test cases. In this intercomparison, 13 radiotherapy institutions in The Netherlands participated. The planning systems evaluated comprise all commercial systems in clinical use in The Netherlands (different versions of in total four manufacturers) as well as in-house developed systems. The five test cases were designed in order to quantify the errors introduced by: (a) the calculation algorithm including the method of source specification used; (b) the localisation procedure via orthogonal radiographs; and (c) the clinical dose specification system in the case of a geometrically irregular implantation for which no established, rigid dose specification method is available. The results of this comparison can be summarized as follows: The accuracy of most calculation algorithms appears to be satisfactory except for the region close to the longitudinal axis of linear sources. Source strength specification in terms of activity or milligram radium equivalent can result in dosimetric errors which may be clinically relevant and therefore specification of the source strength in terms of "output", i.e. air-kerma rate at a specified distance, is to be preferred. This concurs with recent national and international recommendations regarding source strength specification. Localisation via orthogonal radiographs, as commonly used, appears to cause noticeable errors. This aspect could be improved in a quality assurance program. In case of "irregular" implantations, the uncertainty in clinical dose specification clearly dominates the overall accuracy. This illustrates the need for a consistent dose specification system, which is capable of including interstitial implants with a source geometry deviating from the rules given by the existing specification systems. It can be concluded that in brachytherapy, whenever possible, a well-defined, rigid clinical dose specification system should be applied, but that even in such cases the delivered dose may be in error by more than 5% (1 S.D.).