$318 Friday/Saturday, 20-21 September 2002 Posters

Conclusion: EPI quality was found adequate to detect set-up accuracy in square phantom (76 Gy at the isocenter). patients receiving pelvic radiotherapy, however visibility of antero-posterior Four different ROIs simulating the rectum were inserted: three were paral- fields were better than lateral. Although set-up errors were in acceptable lelepipeds positioned at three different distance from the isocentre; one pre- levels in all direction, maximum errors were seen in AP axis. There was no sented a more realistic rectum-like 3D shape. significant change of set-up errors along the treatment. Results and Conclusions: Concerning volume calculation, maximum devia-

tions between different TPS were within 5-6 % and seemed to be mainly due to small digitizing uncertainties. When looking at DVH calculation,

TREATMENT P L A N N I N G A N D D O S E C A L C U L A T I O N S DVHs showed maximum differences in terms of % of rectal volume typical- ly around 5-6 % in the region between 40 and 70 Gy. When considering the average DVHs, the deviations from the mean of the four TPS were gener-

1094 Poster ally within 2-2.5 %. Larger discrepancies (as great as 10 %) were found in Improved commissioning of treatment planning systems the steepest part of the curves, especially near Dmax. The behaviour of R. Jerai 1,2, G. O/ivera 1,2, P. Reckwerdt 2, J. Kapatoes 2, K. Rucha/a 2, each system tended to be the same with every kind of rectum ROI, sug-

gesting that most differences should be systematic. These preliminary W. Lu 2, T.R. Mackie 1,2 results show a good consistency between the systems in calculating rectal 1University of Wisconsin - Madison, Medical Physics, Madison, U.S.A. DVH, apart the region of the DVH very near to Dmax. The uncertainty in rec- 2TomoTherapy Inc., Middleton, U.S.A. tal DVH calculation due to the use of the four different TPS considered in

this work seems to be generally much lower than the uncertainties due to Most of the current dose calculation algorithms in treatment planning sys- contouring uncertainty and set-up/organ motion. Moreover, as the system- tems (TPS) are model-based, like convolution/superposition (C/S). These atic component seems to be dominant, appropriate correction factors for algorithms require several parameters to be set during commissioning, each TPS could further reduce this uncertainty. Unfortunately, dose distributions, which c o m p s - t h e majority of the mea- sured quantities during commissioning, only weakly%-'0~:ret'at~with the input 1097 Poster parameters. Furthermore, current TPS use Monte Carlo kernel distributions Dosimetri¢ verification of ¢onvelution/superposition photon and energy fluence bins, which are very sparse at high energies (over 1 dose calculation models on a commercial treatment planning MeV), but dense at low energies. It will be shown that in order to increase system reliability of the calculations, the commissioning of a TPS should be per- M.M. Aspradakis, R.H. Kermode, N.D. Richmond, A. Stee/e, G.P. Lawrence formed with strong Monte Carlo simulation support. In addition, a superior Newcast/e Genera/Hospita/, Regiona/Medica/Physics Department, New- set of C/S kernels and energy fluence bins will be suggested that increases cast/e upon Tyne, United Kingdom accuracy of the calculation. Two hypothetical linac systems were modeled, with the incident electron Introduction: The accuracy requirements of photon dose calculation algo- energy of 6 MeV but different target thicknesses. The depth-dose curves rithms employed for radiotherapy treatment planning have increased due to were used for commissioning of a C/S based TPS. In addition, the dose was more complex beam set-ups and the routine use of patient density informa- scored according to the incident pencil beam energy. It is interesting that, tion from CT. The criteria of acceptability of dose calculations depend on the while both systems have significantly different photon spectra, the corre- complexity of the irradiation geometry; in most cases tolerance values do spending depth dose curves are remarkably similar. Further study revealed not exceed 3%/3mm. The purpose of this work was to verify the dosimetric that the reason was a similar high-energy part of the spectrum, while not the accuracy with which the pencil beam convolution (TMS-PB) and the col- low energy part. When the C/S code was auto-commissioned, very different lapsed cone superposition (TMS-CC) algorithms on the Helax-TMS plan- spectra were obtained. This indicates that the errors in the spectra were ning system (version 6.1, MDS Nordion) calculate absolute dose in a range "compensated" with other parameters, of homogeneous and heterogeneous density media. Methods: The perfor- It was found out that the commissioning of a model-based TPS could lead mance of the Helax-TMS photon dose calculation models was compared to incorrect input parameters because of the ill-posed nature of the problem, against measurements for 4, 6 and 15 MV. Measurements were carried out This may lead into uncontrolled errors during dose calculation in real either in a water tank or in solid water (WT1) material using ionisation cham- geometries, which were not tested during commissioning. Therefore, a bers and film (Xomat-XV). Irradiation geometries included a range of strong Monte Carlo calculation support to commissioning is suggested. It beams:open and wedged square, rectangular, asymmetric and extended was further found out that the dose characteristics are to a great extent SSD. The heterogeneous phantoms investigated contained air, cork and dominated by the high-energy part of the spectra. Therefore, very different bone heterogeneities. Each algorithm performance was also investigated spectra (up to 25% difference in the mean energy) can produce very simi- under missing tissue irradiation geometries namely oblique incidence, tan- lar dose distributions in homogenous phantoms. Because of the importance gential irradiation and at regions close to beam exit. These geometries were of the high-energy part of the spectrum, the use of the default set of kernels three-dimensionally simulated on the TPS. Calculated data were generated and energy fluence bins for the C/S dose calculation is not adequate. An in terms of absolute dose normalised to the dose under calibration condi- alternative set with the superior dose calculation accuracy was derived, tions (Helax-TMS 'output factor normalised' dose) allowing verification of

monitor units. 1095 Poster Results: Absolute dose differences from measurements for both TMS-PB Comparing rectal DVH calculation with four commercial t reat - and TMS-CC in water with open, wedged, asymmetric and oblique inci- men t planning systems by means of the AAPM TG 23 18 MV dence beams are within the recommended tolerance values. For heteroge- uni t neous and missing tissue irradiation geometries differences from measure- F. Declich 1, L. Masi 2, L. Menegotti 3, M. Stasi 4, C. Fiorino 5 ments up to 10% were observed for the TMS-PB algorithm whilst the results

from the TMS-CC were within recommended tolerance values. 10spedale "A. Manzoni", U.O. Radioterapia, Lecco, Italy Conclusions: This work demonstrates that the TMS-PB model is accurate 2Casa di Cura S.Chiara, U.O. Radiobiologia Cfinica Universit& di Firen- for absolute dose calculations within large media with small involvement of ze, Firenze, Italy non-unit density heterogeneities. Dose accuracy with this model is signifi- 30spedale S.Chiara, U.O. Fisica Sanitaria, Trento, Italy cantly reduced close to phantom boundaries. The TMS-CC algorithm gen- 41RCC - Ordine Mauriziano, U.O. Fisica Sanitaria, Torino, Italy erally provides a better approximation to dose in heterogeneous media and

at media boundaries. 5H.S. Raffaele, U.O. Fisica Sanitaria, Milano, Italy

1098 Poster Introduction: integral DVHs of the rectum calculated by four different com- S tudy o f IMRT dose model inadequacies mercial treatment planning systems (TPS) were compared using AAPM TG- J.E. Bavouth, S.M. Morri// 23 18 MV unit. Goal of this work was to demonstrate the consistency of University of Texas Medica/Branch at Ga/veston, Radiation Onco/ogy Phy- DVH calculation of different systems as a part of an Italian multi-centric clin- sics, Ga/veston, TX, U.S.A. ical study regarding correlations between DVH and observed rectum mor- bidity for patients treated for prostate cancer. This work was conducted Purpose: To design and perform tests that identifies the components of within the AIRO national working group on prostate radiotherapy. IMRT dose distributions which are modeled inaccurately by treatment plan- Materials and methods: The AAPM TG-23 18 MV x-ray unit, previously ning system (TPS) algorithms. implemented and carefully tested on 4 different commercial TPS (Varian Materials and Methods: Doses were computed using two commercially Cadplan, Nucletron Plato, ADAC Pinnacle and Elekta PrecisePlan), was available IMRT TPS's. Calculations and measurements utilized 6 MV x- used for simulating a prostate treatment with the 4-fields technique on a rays passing through a double focus MLC containing 29 leaf pairs (27 pairs