Posters Friday/Saturday, 20-21 September 2002 $319

of 1 cm width). Dosimetric metrics considered were the dose within the NORDION and PLATO, NUCLETRON). Form and size of the PTV were field, in immediately adjacent voxels, 'and in the tails outside of the field, tried to keep as close as possible to the first planning, also machine data Absolute dose and depth doses were measured using MOSFET detectors like gantry angles etc. for field sizes of 1,3, 5, and 10 cm per side, at 5 depths. Dose profiles were Results: Main difficulty was to enter limits and weighting factors to PTV and measured with EDR2 film at 10 cm depth using individual and multiple lx27 risk organs as well as weighting of risk organs relative to the PTV since no cm fields. Both film and MOSFET measurements were verified with a 0.015 standard data (.in many cases) are available: Also dose specification rises cm ~ ionization chamber questions: in some cases specification to ICRU-criteria does not seem to be Results: Depth dose values for small fields were overestimated by both TPS sufficient; evaluation of DVHs has a larger influence. Different ways of algorithms, with relative errors of 6-18% at depths of 15 cm and greater, dosage are offered by different systems. Output factors were overestimated by 2% for TPS1 and 17% for TPS2 when Handling of the systems is also strongly under development. Concepts of l x l cm fields were elongated to lx27 cm. Dose profiles for a single lx27 "normal" 3D conformat planning have to be evaluated; but main difficulties cm strip show penumbra values were within the dose grid resolution, but arise from different modalities to input constraints like total dose, dose to noticeable differences were found outside of the field. TPS1 overestimated parts of a risk organ or weighting over- and underdosed volumes. doses in the region 1 cm beyond the field edge (max relative error 26%) and Planning verification is supported in very different ways, integrated into "rou- underestimated doses in the tails (max relative error 25% 6 cm away from tine modatities" in one system up to the necessity of. "special code inputs" in the field edge). TPS2 model underestimated in both regions, by 40% and others. 7%, respectively. Dose profiles through a bar pattern of 11 strips show the Discussion: As also mentioned by the IMRT Working Group (IJROBP 51(4), ramifications of these differences. TPS1 accurately modeled doses within 880) one main part in future is to standardize not only planning algorithms the field, and overestimated doses between fields (max relative error 42%). but also the different ways of specifying constraints. The actual comparison When using the TPS2 beam model commissioned for conventional radio- is meant as a step on the way to integrate IMRT in routine. therapy, the planning system over-estimated the dose by greater than 36% within the field and 78% in blocked regions between the fields. These errors 1101 Poster were reduced to less than a 10% overestimate within the field and 3% A solution for data and image exchange between a v i r tua l between the fields when beam models were scrutinized and adjusted for s imula tor , treatment planning system and linear accelerator IMRT. from different vendors Conclusions: Knowledge of the conditions in which IMRT TPS algorithms Y. van Herren, J. Wiersma, R. van Os, G. D'O/ies/ager, C. Koedooder fail allows users to optimize their TPS model and utilize MLC segmentation Academic Medical Center, Radiotherapy, Amsterdam, The Netherlands algorithms to minimize these dose errors in clinical IMRT treatments.

1099 Poster Most vendors on the radiotherapy (RT) marketplace claim their equipment nowadays to be compliant to the Dicom-RT standard for data interchange.

Forward-planned dose compensation for radiotherapy of the However, the implementation of a network for data and image exchange breast using MLC sub- f ie lds between RT equipment of different vendors often needs solutions, tailored B. Nyiri, L. Geriq, M. MacPherson, J. Szanto to the local situation on a RT department. In our institute, the need for elec- Ottawa Regional Cancer Centre, Medical Physics, Ottawa, Canada tronic data transfer, already being encouraged for efficiency reasons, was

still enhanced by the hospital strategy towards working filmiess and paper- Many patients who undergo radiotherapy of the breast will develop clinical- less. In this presentation we report on our experiences with establishing ly significant sequelae as a result of their radiation treatment, including electronic exchange of RT-data and images between a Virtual Simulator acute radiation dermatitis, fibrosis, telangiectasia, mastalgia, and peau d'o- (AcQsim, Marconi/Philips), a Treatment Planning System (Plato, Nucle- range. Many of these complications result from dose inhomogeneity due tron), linear accelerators (SL-15i, Elekta), a Portal Imaging system to the complex shape Of the breast and consequent non-uniformity of atten- (iView/SRl-100, Elekta) and a database from which we print treatment uating tissue. Some institutions have employed IMRT techniques to charts. Recently our Oldelft conventional simulator was equipped with DTI improve the dose uniformity within the breast. However, the inverse plan- software for digitizing conventional simulator images, making also this sys- ning algorithms required for this approach are not available at many con- tern a suitable rink for image transfer. Commercial software as well as pack- tres. Missing tissue compensators can also improve the dose uniformity, ages developed at the department or i n cooperation with Nucletron were but are time-consuming in terms of both manufacture and treatment deliv- implemented. Differences between conventionally and electronically han- ery. In this work, we have developed a technique for approximating a con- died data and image flows will be presented, together with the gain in effi- tinuous missing tissue compensator, designed on a conventional 3D radio- ciency and altered working methods. E.g., the Elekta MLP workstation, for therapy treatment planning system (TheraPlan Plus, MDS Nordion, Ottawa, definition of multileaf prescriptions for the SLi machines and the production Canada), using MLC sub-fields. The algorithm is based on the least of reference images for iView, can be avoided by sending the AcQsim squares approximation of the fluence distribution leaving the compensator DRR's parallel with the multileaf settings from Plato or AcQsim directly and shield/block system by a multi-step function. The number of sub-fields through the network to the treatment machines. is chosen by the user, and the field shapes and weights are calculated to provide the minimum deviation from the continuous fluence distributio n. 1102 Poster Tests of a conventional medial-lateral beam arrangement on an anthropo- Experimental validation of FFT convolution and mul t ig r id morphic phantom show that an open-beam hot spot of 24% can be reduced superpos iUon a lgo r i t hms for low density media dose calcula- to 8% using four sub-fields per beam, and 6% using 6 sub-fields. This corn- t ions pares with a hot spot of 7% using a physical compensator. This technique is applicablein any centre equipped with multileaf collimators and the abili- F. Garcfa-Vicente, A. Mi~ambres, I. Jerez-Sainz, I. Modolell, L. P~rez- ty to design dose or missing tissue compensators. Gonzdlez, J.J. Tortes

Hospital la Princesa, Radiotherapy and Medical Physics, Madrid, Spain

1100 Poster Purpose: Conformal radiotherapy requires accurate dose calculation in

IMRT - A look at the present status of treatment planning sys- each relevant clinical situation. terns One of the most critical situations of the dose calculation by the treatment W.F.O. Schmidt, W. Nespor, K. Pay/as, R. Haw/iczek planning system (TPS) could be the radiotherapy treatment of lung cancer /nst f Radioonco/ogy, Donauspita/Vienna, Vienna, Austria tumour. The aim of this paper is to assess the calculation accuracy in het-

erogeneous tissues of two modern physics-driven algorithms, Fast Fourier Introduction: Since 1999 IMRT patients are irradiated at the Institute for Transform convolution (FFTC) and MultiGrid Superposition (MGS), both Radiooncology, a routine department in the Donauspital Vienna, with a implemented in a commercial TPS. SIEMENS KD2 linac. Treatment planning is performed with a CORVUS- Methods and Materials: It was investigated the calculation accuracy of the system (NOMOS), which was exclusively developed for IMRT and offers FFTC and MGS algorithms implemented in the FOCUS 3.0.0 TPS system special features eg for designing and weighting of risk organs, input of (Computerized Medical Systems, USA). Both algorithms were previously uncertainties due to immobilization or organ movement or verification of validated in water according to Van Dyk criteria. To evaluate them, two treatment plans. Since now (Apr 2002) three other centers in Vienna also phantoms simulating different configurations, chest wall-lung-chest wall and perform IMRT (with different linacs and planning systems) and a fourth is mediastinum-lung, have been irradiated using photon beams of 6 and 25 starting soon, so a trial to "compare" systems was undertaken. MV nominal energy from a Satume43 lineal accelerator (GE Medical Sys- Methods and materials: Different sets of typical patient data (mainly H&N) tern). The calculated data were.compared with measured data, using ion- who were already planned and treated at the Donauspital were read into ization chamber and radiographic film, through percentage depth dose four other planning systems (FOCUS, CMS; HELIOS, VARIAN; TMS, MDS-