Assessment of the uncertainties in dose delivery of a commercial system for linac-based stereotactic radiosurgery

Purpose: Linac-based stereotactic radiosurgery (SRS) was introduced in our department in 1992, and since then, more than 200 patients have been treate d with this method. An in-house-developed algorithm for target localization and dose calculation has recently been replaced with a commercially availa ble system. In this study, both systems have been compared, and positional accuracy, as well as dose calculation, have been verified experimentally. Methods and Materials: The in-house-developed software for target localizat ion and dose calculation is an extension to George Sherouse's GRATIS(R) sof tware for radiotherapy treatment planning, and has been replaced by a comme rcial (BrainSCAN version 3.1; BrainLAB, Germany) treatment planning system (TPS) for SRS, The positional accuracy for the entire SRS procedure (from i mage acquisition to treatment) has been investigated by treatment of simula ted targets in the form of 0.2-cm lead beads inserted into an anthropomorph ic phantom. Both dose calculation algorithms have been verified against man ual calculations (based on basic beam data and CT data from phantom and pat ients), and measurements with the anthropomorphic phantom applying ionizati on chamber, thermoluminescent detectors, and radiographic film. This analys is has been performed on a variety of experimental situations, starting wit h static beams and simple one-are treatments, to more complex and clinical relevant applications. Finally, 11 patients have been evaluated with both T PS in parallel for comparison and continuity of clinical experience, Results: Phantom studies evaluating the entire SRS procedure have shown tha t a target, localized by CT, can be irradiated with a positional accuracy o f 0.08 cm in any direction with 95% confidence. Neglecting the influence of dose perturbation when the beam passes through bone tissue or air cavities , the calculated dose values obtained from both TPSs agreed within 1% (SD 1 %) for phantom and patient studies. The application of a one-dimensional pa th length correction for tissue heterogeneity influences the treatment pres cription 4% on average (SD 1%), which is in compliance with theoretical pre dictions, The phantom measurements confirmed the predicted dose at isocente r within uncertainty for the different treatment schedules in this study. Conclusion: The full SRS procedure applied to an anthropomorphic phantom ha s been used as a comprehensive method to assess the uncertainties involved in dose delivery and target positioning. The results obtained with both TPS s are in agreement with AAPM Report 54, TG 42 and clinical continuity is as sured. However, the use of a one-dimensional path length correction will re sult in an increase of 4% in dose prescription, which is slightly more than that predicted in the literature. (C) 1999 Elsevier Science Inc.