- Email: [email protected]
105 oral Development and validation of a multiple source model for radiosurgery
$42
Tuesday, 16 September 2003
Proffered papers DOSE C O M P U T A T I O N 104
oral
The e f f e c t of tissue inhomogeneity in dose based and biologically b a s e d I M R T N. Papanikolaou, L. Gutierrez, Y. Yan, C. Wu, J. Penagaricano, V. Ratanatharathom University of Arkansas, Radiation Oncology, Little Rock, Arkansas, U.S.A. In the new era of IMRT the importance of accurate close calculation is further elevated. The goal of IMRT is to produce a plan that will deliver optimum conformance to the tumor. If the effects of tissue inhomogeneities are ignored or miss-interpreted during the calculation, then the deliverable distribution is not the optimal one. In this work we investigated the effect of tissue inhomogeneity on IMRT planning for some clinically relevant cases. Plans were optimized using 6x only and 18x only photon beams (coplanar), in order to study not only the effect of different types of inhomogeneity algorithms on the IMRT plan, but also to investigate the magnitude and significance of the effect as the energy of the beam changes. In the study, we compared homogeneous dose calculation to pencil beam, convolution and superposition. The effect of the dose algorithm was evaluated on both Dose optimization and Biological optimization plans. The dose optimization was based on dose and dose volume objectives, while the biological optimization used the TCP/NTCP, EUD and P+ models. Our results showed several differences between the homogeneous calculation and the more advanced algorithms for both energies. The absolute dose was found to be several percent units (up to 20%) different proximal to tissue interfaces, especially tissue-bone and air-tissue. The location and value of the hot spots can also be quite different. Plans were compared based on isodose distribution coverage and on DVHs. Based on our findings we concluded that for IMRT, tissue inhomogeneities should be included in the optimization, or at least be computed both with and without for a more realistic clinical assessment. 105
oral
Development and v a l i d a t i o n o f a m u l t i p l e source model for radiosurgery A. Chaves 1, C. Oliveira2, M.C. Lopes 1, L. Peralta 3 1Centro Regional Oncologia Coimbra, Radiotherapy, Coimbra, Portugal 21nstituto TecnolSgico e Nuclear, Sacav~m, Portugal 3LaboratSdo de Instrumenta~&o e Ffsica Experimental de Partfculas, Lisboa, Portugal Introduction: In the past few years, MC simulations appear as powerful tool to access photon beam characterization in radiotherapy. These detailed studies have lead to the development of multiple source models. These models are based on the fact that particles originated in the same component of the accelerator head have similar distributions: position, direction, energy, weight. A reconstructed radiation beam is then composed by particles from various virtual sources representing each relevant component of the accelerator head. The application of this idea to radiosurgery narrow photon beams seems attractive since the reduction in time and data storage is particularly important. The goal of this work is to develop a multiple source model for radiosurgery narrow photon beams. The model will be validated through comparisons with measurements in a water phantom. Methods: MCNP4C code was used to produce the phase space data (PSD) of each additional collimator. Eight photon relevant virtual sources were extracted from an extensive characterization of these PSDs. These virtual sources were introduced in the very fast MC code DPM for close calculations in a water phantom. The size of the voxels in water was fixed to 1mm x lmm x 5mm. Results: Calculated depth dose curves and profiles were compared with measurements in water for all the additional collimators. Results are within accepted international tolerances. Conclusioq: A multiple source model applied to radiosurgery narrow photon
beams was developed and validated through measurements in water. This model will be the basis of a MC radiosurgery dose calculation engine. 106
oral
The necessity o f a d i v e r g e n t initial electron beam in the Monte-Carlo model of a linear accelerator B. De Smedt 1, N. Reynaert 1, M. Coghe2, L. Paelinck2, B. Van Duysse 2, C. De Wagtef 2, W. De Neve2, H. Thierens 1 I Gent University, Medical Physics Dept., Gent, Belgium 2Gent University Hospital, Radiotherapy Dept., Gent, Belgium In the modelling process of the Elekta SL18 linear accelerator of the Ghent University Hospital, for which results of MLC modelling have been published previously, the electron source configuration was investigated thoroughly. The Linac is modelled accurately in BEAMnrc and the obtained phasespace files are used as input in Dosxyznrc to determine depth dose distributfons and lateral profiles in a water phantom. Measurements and calculations were performed with the flattening filter and difference filter (for 18 MV) removed to eliminate their effect. For spot optimisation 40x40 beams were used to limit the effect of the collimating devices on the flatness of the profiles. Especially for 18 MV it was clear that the agreement between measurement and calculation was rather poor. Including a small divergence angle to the initial beam solved this issue. This was simulated by altering the electron focus position (which is not focussed perfectly on the target) and allowing a maximal divergence of the electron beam of 2 ° before hitting the target. A second important issue concerning the flatness of the lateral profiles, is the BEAM energy. We started from a measured electron spec* trum and optimised it to obtain a perfect agreement with the measurements. For 6 MV the agreement was obtained straight away. Again for 18 MV we had more problems and the electron energy had to be decreased. Some details of the difference filter, which has a complicated geometry, were very important and needed to be modelled in great detail. To obtain a good agreement in the penumbra region the spot size had to be optimised. Conclusion: We are planning to insert a beam model based on the obtained phase-space files in our MC treatment planning system, therefore a good starting point is necessary and attention should be paid to the modelling of the accelerator. 107 oral V a l i d a t i o n o f m o n i t o r u n i t c a l c u l a t i o n s at arbitrary source-to-
surface distance for the new commercial Monte Carlo based treatment planning system for electron beams J. Cvcller, G. Daskalov Ottawa Regional Cancer Centre, Medical Physics, Ottawa, Canada Purp0_se: Monitor unit calculations for electron beams at arbitrary SSDs are known to be very inaccurate with existing treatment planning systems. Typically users are faced with creating separate virtual machines for different SSDs to circumvent problems related to the inabiflty of the treatment planning software to handle the inverse square law corrections. In the recently released MDS Nordion electron Monte Carlo treatment planning module (DCM) in TheraplanmM Plus users need to create only one virtual machine for each electron energy to be able to handle MU calculations for arbitrary SSDs. In this paper we present the verification of monitor unit calculations for this software at the standard SSD=100 cm and the extended SSD=115 cm distances most often used in our clinic. Methods: The verification tests are performed for all applicators and beam energies 6-20 MeV available on a Siemens KD2 accelerator. Sets of 9 rectangular and 17 square cutouts are used for each energy. DCM is used to calculate the monitor unit required to deliver a prescribed dose at depth dmax along the beams central axes in a homogeneous water phantom. 50,000 histories per cm 2 are used in each calculation to achieve relative statistical uncertainties of about 1%. DCM results are verified by comparison with manual calculations based on a comprehensive set of experimental data. Results: The results of Monte Carlo and hand calculations for each SSD are
Tuesday, 16 September 2003
Proffered papers DOSE C O M P U T A T I O N 104
oral
The e f f e c t of tissue inhomogeneity in dose based and biologically b a s e d I M R T N. Papanikolaou, L. Gutierrez, Y. Yan, C. Wu, J. Penagaricano, V. Ratanatharathom University of Arkansas, Radiation Oncology, Little Rock, Arkansas, U.S.A. In the new era of IMRT the importance of accurate close calculation is further elevated. The goal of IMRT is to produce a plan that will deliver optimum conformance to the tumor. If the effects of tissue inhomogeneities are ignored or miss-interpreted during the calculation, then the deliverable distribution is not the optimal one. In this work we investigated the effect of tissue inhomogeneity on IMRT planning for some clinically relevant cases. Plans were optimized using 6x only and 18x only photon beams (coplanar), in order to study not only the effect of different types of inhomogeneity algorithms on the IMRT plan, but also to investigate the magnitude and significance of the effect as the energy of the beam changes. In the study, we compared homogeneous dose calculation to pencil beam, convolution and superposition. The effect of the dose algorithm was evaluated on both Dose optimization and Biological optimization plans. The dose optimization was based on dose and dose volume objectives, while the biological optimization used the TCP/NTCP, EUD and P+ models. Our results showed several differences between the homogeneous calculation and the more advanced algorithms for both energies. The absolute dose was found to be several percent units (up to 20%) different proximal to tissue interfaces, especially tissue-bone and air-tissue. The location and value of the hot spots can also be quite different. Plans were compared based on isodose distribution coverage and on DVHs. Based on our findings we concluded that for IMRT, tissue inhomogeneities should be included in the optimization, or at least be computed both with and without for a more realistic clinical assessment. 105
oral
Development and v a l i d a t i o n o f a m u l t i p l e source model for radiosurgery A. Chaves 1, C. Oliveira2, M.C. Lopes 1, L. Peralta 3 1Centro Regional Oncologia Coimbra, Radiotherapy, Coimbra, Portugal 21nstituto TecnolSgico e Nuclear, Sacav~m, Portugal 3LaboratSdo de Instrumenta~&o e Ffsica Experimental de Partfculas, Lisboa, Portugal Introduction: In the past few years, MC simulations appear as powerful tool to access photon beam characterization in radiotherapy. These detailed studies have lead to the development of multiple source models. These models are based on the fact that particles originated in the same component of the accelerator head have similar distributions: position, direction, energy, weight. A reconstructed radiation beam is then composed by particles from various virtual sources representing each relevant component of the accelerator head. The application of this idea to radiosurgery narrow photon beams seems attractive since the reduction in time and data storage is particularly important. The goal of this work is to develop a multiple source model for radiosurgery narrow photon beams. The model will be validated through comparisons with measurements in a water phantom. Methods: MCNP4C code was used to produce the phase space data (PSD) of each additional collimator. Eight photon relevant virtual sources were extracted from an extensive characterization of these PSDs. These virtual sources were introduced in the very fast MC code DPM for close calculations in a water phantom. The size of the voxels in water was fixed to 1mm x lmm x 5mm. Results: Calculated depth dose curves and profiles were compared with measurements in water for all the additional collimators. Results are within accepted international tolerances. Conclusioq: A multiple source model applied to radiosurgery narrow photon
beams was developed and validated through measurements in water. This model will be the basis of a MC radiosurgery dose calculation engine. 106
oral
The necessity o f a d i v e r g e n t initial electron beam in the Monte-Carlo model of a linear accelerator B. De Smedt 1, N. Reynaert 1, M. Coghe2, L. Paelinck2, B. Van Duysse 2, C. De Wagtef 2, W. De Neve2, H. Thierens 1 I Gent University, Medical Physics Dept., Gent, Belgium 2Gent University Hospital, Radiotherapy Dept., Gent, Belgium In the modelling process of the Elekta SL18 linear accelerator of the Ghent University Hospital, for which results of MLC modelling have been published previously, the electron source configuration was investigated thoroughly. The Linac is modelled accurately in BEAMnrc and the obtained phasespace files are used as input in Dosxyznrc to determine depth dose distributfons and lateral profiles in a water phantom. Measurements and calculations were performed with the flattening filter and difference filter (for 18 MV) removed to eliminate their effect. For spot optimisation 40x40 beams were used to limit the effect of the collimating devices on the flatness of the profiles. Especially for 18 MV it was clear that the agreement between measurement and calculation was rather poor. Including a small divergence angle to the initial beam solved this issue. This was simulated by altering the electron focus position (which is not focussed perfectly on the target) and allowing a maximal divergence of the electron beam of 2 ° before hitting the target. A second important issue concerning the flatness of the lateral profiles, is the BEAM energy. We started from a measured electron spec* trum and optimised it to obtain a perfect agreement with the measurements. For 6 MV the agreement was obtained straight away. Again for 18 MV we had more problems and the electron energy had to be decreased. Some details of the difference filter, which has a complicated geometry, were very important and needed to be modelled in great detail. To obtain a good agreement in the penumbra region the spot size had to be optimised. Conclusion: We are planning to insert a beam model based on the obtained phase-space files in our MC treatment planning system, therefore a good starting point is necessary and attention should be paid to the modelling of the accelerator. 107 oral V a l i d a t i o n o f m o n i t o r u n i t c a l c u l a t i o n s at arbitrary source-to-
surface distance for the new commercial Monte Carlo based treatment planning system for electron beams J. Cvcller, G. Daskalov Ottawa Regional Cancer Centre, Medical Physics, Ottawa, Canada Purp0_se: Monitor unit calculations for electron beams at arbitrary SSDs are known to be very inaccurate with existing treatment planning systems. Typically users are faced with creating separate virtual machines for different SSDs to circumvent problems related to the inabiflty of the treatment planning software to handle the inverse square law corrections. In the recently released MDS Nordion electron Monte Carlo treatment planning module (DCM) in TheraplanmM Plus users need to create only one virtual machine for each electron energy to be able to handle MU calculations for arbitrary SSDs. In this paper we present the verification of monitor unit calculations for this software at the standard SSD=100 cm and the extended SSD=115 cm distances most often used in our clinic. Methods: The verification tests are performed for all applicators and beam energies 6-20 MeV available on a Siemens KD2 accelerator. Sets of 9 rectangular and 17 square cutouts are used for each energy. DCM is used to calculate the monitor unit required to deliver a prescribed dose at depth dmax along the beams central axes in a homogeneous water phantom. 50,000 histories per cm 2 are used in each calculation to achieve relative statistical uncertainties of about 1%. DCM results are verified by comparison with manual calculations based on a comprehensive set of experimental data. Results: The results of Monte Carlo and hand calculations for each SSD are