E. Panitsa¹, J. C. Rosenwald², C. Kappas¹
1. University of Patras, Medical Physics Department, Patras (Greece)
2. Institut Curie, Service de Physique Medicale, Paris (France)

Advanced techniques in radiotherapy make extensive use of computers for precise planning of the irradiation. It is therefore mandatory to have access 10 TPS which have adequate performances and are fully validated. A cooperative work has been undertaken on this subject within thee framework of thee E.C. BIOMED 1 project DYNARAD. A methodology for the Quality Control of Treatment Planning Systems consisting of three classes of tests (i.e. qualitative tests, internal validation, and quantitative tests of the sophisticated dose calculation features) has been set up. The application of the methodology started with the development of a series of tests for qualitative assessment of dose calculation features such as irregular field, lack of tissue, 2D and 3D inhomogeneity corrections, tray contamination. The protocol was distributed 10 the DYNARAD participants and results from 8 different TPS were obtained. The results showed some limitations of the systems.

TPS which are presently the main computerized tool at the design of the treatment plan in radiotherapy, are getting more and more sophisticated as they integrate advanced features for the realization of modem techniques of conformal radiotherapy. The increasing sophistication results in an increased risk of error. Moreover, the consequences of such errors are of greater importance than in the conventional approach [i]. It is therefore, mandatory to have fully validated TPS with adequate performance in all cases used clinically

A working team in the frame of the European project DYNARAD has decided to develop a protocol on Quality Control (QC) of Treatment Planning Systems (TPS), with emphasis to their three dimensional (3D) characteristics. DYNARAD is a three year concerted action project supported by the European Commission, in which 29 Institutions from 10 European countries combine their efforts in the field of "Development and Standardization of New Dynamic Radiotherapy Technics".

A methodology was developed 12-41 in order to check ail the TPS features in terms of performance and accuracy. The proposed methodology was based on three classes of tests with the following characteristics:

• Class I : Qualitative assessment of the general features of TPS.

This assessment requires no experimental data. It is a strictly qualitative assessment that examines only if the TPS accounts for a specific feature and not how accurately it performs.

• Class 2: Internal validation of TPS using "home" beam data

These tests are to determine whether the TPS calculations are consistent with home beam measurements. A list of simple standard situations [5] bas to be decided and tested. Experimental data are required.

• Class 3. Quantitative assessment of the sophisticated dose calculation features

Quantitative tests should also be performed for the more sophisticated situations. Experimental data is required that is sometimes difficult to extract.

A test protocol containing a series of test cases assessing qualitatively various dose calculation features, was prepared and distributed to the DYNARAD participants, as the first step in the "Quality Control of TPS" [6]. The proposed tests belong to the class I "qualitative assessment" category. These tests are easy to perform by any physicist wishing to check whether its TPS really corporate specific dose calculation features. The tests use the home beam data library. No additional measurements are required. The dose calculation features were selected as the first ones to be included to the protocol, since they are the most crucial characteristics of a TPS. The selected test cases were mostly drawn from the literature [1,7-10], paying special attention to those features which are believed to be weak point of the presently used dose calculation algorithms. The features assessed by the proposed test cases presented in fig. 1, are the following:

· Surface dose and electron contamination : The surface dose was calculated for an open field, a centrally blocked field, and a field with a tray (fig. 1a,b,e). The comparison of the surface dose values with and without a tray (or a block) showed if the TPS is taking into account the electron contamination. The comparison of the surface dose values on axis and at 5, - 5 cm off axis showed whether the surface dose value depends on the position of the calculation point.

· Block correction: The dose distribution was calculated for a centrally placed block of dimensions: 3 x 3, 3 x 25, and 7 x 25 cni² (fig. 1b). The comparison between them shows if the dose calculation algorithm takes into account the block both laterally and longitudinally.

· Inhomogeneity correction (within, below, lateral, or 3-D): The dose distribution was calculated for a phantom with an inhomogeneity placed consecutively on the beam axis, off-axis or off-plane (fig. 1c). The comparison with the reference case of a homogeneous phantom determines the limits of the inhomogeneity correction algorithm.

· Inhomogeneity correction below a block : The dose distribution was calculated for a combination of a phantom having an horizontal inhomogeneity and a centrally blocked field (fig. 1f). The comparison with the case of the homogeneous phantom having the same beam configuration, showed whether the inhomogeneity correction is taken into account when a block is present.

· Lack of scattering, material (lateral, T-shape, inverse T-shape, 3D): The dose distribution was calculated for a phantom lacking material laterally, with T-shape and inverse T-shape configurations, and at a different plane (fig. Id). The comparison with the reference case of the cubic phantom showed the ability of the TPS to take account of the lack of scattering material.

The tests were performed by nine institutions participating in the Dynarad project. The following eight TPS's were evaluated:

1. PLATO RTS v I .3 (Acadeinisch Medisch Centrum, Amsterdam, Holland)

2. MODULEX 3000 CMS v2.75 (Borgo Trento Hospital, Verona, Italy)

3. CADPLAN Varian Dosetek v2.62 (University Central Hospital, Helsinki, Finland)

4. HELAX TMS v2.10H (Arhus Kommune hospital, Aarhus, Denmark and Haukeland Hospital, Bergen, Norway)

5. DOSIGRAY v95 (Institut Gustave Roussy, Villejuif, France)

6. CARO (home made) (Instituto Regina Elena, Rome, Italy)

7. MEVAPLAN (University Hospital, Patras, Greece)

8. ISIS-2 V2.0 (Institut Curie, Paris, France)

The performance of each evaluated TPS is presented in Table I. A brief description of the results [4] for every assessed feature is as follows:

• Surface dose and electron contamination: No TPS was found to take account of electron contamination. Some TPSs do not supply doses for depths in the build-up region, some give a single surface dose value for every point on the surface no matter its x, y coordinates while some supply a surface dose value characterized by the point coordinates.

• Block correction: All examined TPSs are taking account of the block both for the lateral and longitudinal dimension.

• Inhomogeneity correction: All examined TPS correct their dose calculation for points lying within or below an inhoniogeneity. Some TPS correct the dose calculation at points lateral of the inhomogeneity, while none of the examined TPS had any correction for the presence of an inhomogeneity located off-plane. In one case it was demonstrated that the TPS is able to include a 3D inhomogeneity correction, but only if, in the case of lateral inhoinogeneity, a sagittal plane is introduced to perform the dose calculations.

• Inhomogeneity correction below a block: Two of the examined systems included an inhomogeneity correction for points lying below the block.

• Lack of scattering material : Some TPS include a correction in the dose calculation near the edge of the phantom with the exception of the electron disequilibrium region. however, not ail of the TPS that take account of the lateral lack of material, include a correction at the dose calculation for the T-shape phantoms. Obvious artifacts were found in the lack of material case (especially in the T-shape phantom) in two of the examined TPS. Only two of the examined TPS take account of the 3D lack of material.

The examined TPS were of a wide variety. Old and modem systems, two dimensional or with three dimensional capabilities, home-made and commercial ones, were included in our tests. As it was expected, a wide range of performances was found. In brief, the following results were taken:

The results showed the expected deficiences of the sytems to give satisfactory results in the more implicated situations of 3D geometries. Their deficiency was in many cases expanded to 2D geometries like lateral inhomogeneities or lack of material. However, even the more modem systems showed limitations in the more complicated cases. system seems to include 3D inhomogeneity correction when the inhomogeneity is described in cross section distant from the calculation plane. As consequence, it was proved that the commercial implementation of sophisticated algorithms can include simplifications that limit the inherent possibilities of these systens.

The expansion of the protocol to quantitative tests (i.e. classes 2, and 3 of the proposed methodology) is a future task for the DYNARAD Treatment Planning Group. Another task, being presently under development, is the qualitative test of other features, such as no coplanar beams and dose volume histograms.

We express our thanks to the European Commission supporting our work in DYNARAD concerted action, through the BIOMED I program. Special thanks are also given to the physicists K. Koedooder, G. Chierego, T. Hamalainen, J. J. Christensen and M. S.Thomsen, B. Nygaad, J. Chavaudra and H. Beaudre and G. Nicolazic, M. Benassi and L. Begnozzi, C. Dimou, G. Robineau and M. C. Murillo of the DYNARAD Institutions that performed our test protocol.

REFERENCES

1. Rosenwald JC, Drouard J. Quality assurance of 3D treatment planning systems. 3D Treatment Planning, EAR Workshop on three dimensional treatment planning. Three-Dimensional Treatment Planning, WHO Geneva, Oct. 1992;3il-8.
2. Rosenwald JC. Guidelines for Quality Control of 3D Treatment Planning Systens used in Radiotherapy. Dynarad project, Treatment Planning Group: Internal report. 1995.
3. Rosenwald JC. Quality Control of Treatment Planning Systems: Proposal for a methodology. ESTRO, Biennial Meeting on Physics in Clinical Radiotherapy. Gardone Riviera, Italy 8-il Oct 1995. Abstract. Radiotli Oncol 1995: 37(Suppl. 1); 25.
4. Rosenwald JC. Quality Control of 3D Treatment Planning Systems used for Conformal Radiotherapy. Rontgen Centenary Congress, Wurzburg, 20-23 Sep. 1995. Abstract. Medizinische Physik 1995: 95; 168-9.
5. Van Dyk J et al. Commissioning and quality assurance of treatment planning computers. Int J Radiat Oncol Biol Phys 1993: 26; 261-73.
6. DYNARAD, Treatment Planning Group. Report on "Quality Control of 3D Treatment Planning Systems": Qualitative assessment, dose calculation characteristics.
7. Kosunen A et al. Intercomparison of radiotherapy treatment planning systems for external photon and electron beam dose calculation. Radioth Oncol 1993: 29; 327-35.
8. Masterson ME et al. Interinstitutional experience in verification of external photon dose calculation. Int J Radiat Oncol Biol Phys 1991: 21; 37-58.
9. Simonian S, ZeIkili S. Résultats de quelques tests de qualité concernant le calcul de la dose en 3D. 34eme congres SFPH. Strasbourg 1995.
10. Kappas K and Rosenwald JC. Quality control of inhomogeneity correction algorithms used in treatment planning systems. int J Radiat Oncol Biol Phys 1995: 32; 847-58.