- Email: [email protected]
Monte Carlo characterization of materials for prosthetic implants and dosimetric validation of Pinnacle3 TPS
Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Monte Carlo characterization of materials for prosthetic implants and dosimetric validation of Pinnacle3 TPS q Francesca Palleri a,*, Fabio Baruffaldi b, Anna Lisa Angelini a, Andrea Ferri a,1, Emiliano Spezi a,2 a b
Servizio di Fisica Sanitaria, Policlinico S.Orsola-Malpighi, Via Massarenti 9, 40138 Bologna, Italy Laboratorio di Tecnologia Medica, Istituti Ortopedici Rizzoli, via di Barbiano 1/10, 40136 Bologna, Italy
a r t i c l e
i n f o
Article history: Received 11 June 2008 Received in revised form 26 August 2008 Available online 9 September 2008 Keywords: Hip prosthesis Dose calculation Treatment planning system BEAMnrc Monte Carlo code
a b s t r a c t In external beam radiotherapy the calculation of dose distribution for patients with hip prostheses is critical. Metallic implants not only degrade the image quality but also perturb the dose distribution. Conventional treatment planning systems do not accurately account for high-Z prosthetic implants heterogeneities, especially at interfaces. The materials studied in this work have been chosen on the basis of a statistical investigation on the hip prostheses implanted in 70 medical centres. The first aim of this study is a systematic characterization of materials used for hip prostheses, and it has been provided by BEAMnrc Monte Carlo code. The second aim is to evaluate the capabilities of a specific treatment planning system, Pinnacle3, when dealing with dose calculations in presence of metals, also close to the regions of high-Z gradients. In both cases it has been carried out an accurate comparison versus experimental measurements for two clinical photon beam energies (6 MV and 18 MV) and for two experimental sets-up: metallic cylinders inserted in a water phantom and in a specifically built PMMA slab. Our results show an agreement within 2% between experiments and MC simulations. TPS calculations agree with experiments within 3%. Ó 2008 Elsevier B.V. All rights reserved.
1. Introduction The age of population is growing and more patients with hip prostheses are undergoing external beam radiotherapy for pelvic malignancies. Dose planning for patients with metallic implants is challenging because high atomic number prostheses influence heavily the dose distribution prescribed to the target and to the surrounding tissues. The dosimetric aspects of external radiotherapy of patients with hip prostheses have been reviewed and special beam arrangements, in which prosthesis must preferably be avoided, have been advised [1]. Nevertheless, avoiding the hip prosthesis at the time of planning it is not always possible, because it may compromise the dose sparing of the rectum and the bladder. In external beam radiotherapy a detailed knowledge of the dose perturbation introduced by a high-Z prosthetic implant is a very q Part of this work has been presented to the First European Workshop on Monte Carlo Treatment Planning (Ghent 2006). * Corresponding author. Tel.: +39 0516363133; fax: +39 0516363571. E-mail addresses: [email protected], [email protected] (F. Palleri). 1 Present address: Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori (I.R.S.T.), Via Maroncelli 40, 47014 Meldola (FC), Italy. 2 Present address: Department of Medical Physics, Velindre Cancer Centre, CF142TL Cardiff (UK).
0168-583X/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2008.08.013
important treatment planning resource for the correct calculation of the dose distribution delivered to patients. Therefore, knowledge of capabilities and limitations of the treatment planning system in these situations is required. Roberts [2] pointed out the importance of evaluating the accuracy of the dose calculation carried out with a conventional treatment planning system (TPS) by recording dose profiles under a metallic rod immersed in water and by comparing them with calculated profiles. Monte Carlo (MC) methods allow dosimetric characteristics of high-density materials to be evaluated. Many studies deal with MC dose calculation in presence of hip prostheses. In the most recent works there are comparisons between dose calculations carried out with MC methods and with the most common algorithms implemented in TPS [3–7]. Nevertheless these works did not validate MC simulations or TPS calculations with experimental measurements. Recently Buffard et al. [8,9] investigated MC dose calculation for the irradiation of metal alloys; in particular they studied the interface effects, which are local dose perturbations in presence of high-Z gradients. However, these works did not study the effect of these materials at more common photon energies used in radiotherapy treatment planning. In this work we evaluated the dosimetric characteristics of some of the most common materials used for prosthetic implants. We carried out an accurate research to study the characteristics of the prosthetic materials actually implanted in the centres included
5002
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
in this study. The validation of our MC models versus experimental data for clinical 6 MV and 18 MV photon beams was carried out in a earlier study [10]. However, in that study we did not validate TPS calculations against measurements or MC calculations. In the present work we evaluated the collapsed-cones convolution (CC) algorithm employed in Pinnacle3 TPS (Philips, The Netherlands) in presence of high-density materials through the comparison with experimental data, and we focused on the way CC algorithm accounts for interface effects. 2. Materials and methods
Table 2 Chemical compositions of the most common prosthetic materials Titanium alloys
CoCr alloys
Steel alloys
Element
Percent
Element
Percent
Element
Percent
Ti Al V Fe O C N H
Balanced 5.5–6.8 3.5–4.5 0.3 max 0.2 max 0.06 max 0.05 max 0.015 max
Co Cr Mo Ni Fe Mn Si C N
Balanced 26.5–30.0 4.5–7.0 0.1 max 0.1 max 0.1 max 0.1 max 0.1 max 0.25 max
Fe Cr Ni Mn Mo Si N Cu C
Balanced 17.0–19.0 13.0–16.0 2.0 max 2.3–4.2 1.0 max 0.1–0.2 max 0.5 0.03
2.1. Experimental measurements The materials studied in this work have been chosen on the basis of the regional register of orthopaedic prosthetic implants (http://ripo.cineca.it). The 80% of stems and 70% of acetabular cups of the hip prostheses implanted between 2000 and 2004 were manufactured with six materials: two types of titanium alloy, two types of steel alloy, CoCr alloy and ultra high molecular weight polyethylene (UHMWPE). The full list of materials is given in Table 1, with density, international standard code and international trade name. The chemical composition, as provided by the manufacturers, is given in Table 2. We carried out a MC model for these materials, which have been produced in cylinders of 15 cm length and 3 cm diameter for this investigation. For our experiment we used two configurations: (1) the cylinders were immersed in a water phantom (MP3/PTW, Freiburg, Germany). Dose profiles at different depths under the cylinders were acquired with a semiflex ion chamber (type 31010, PTW, Freiburg, Germany) with a sensitive volume of 0.125 cm3. The experimental set-up is shown in Fig. 1; (2) the cylinders were inserted in TRILOGY, a 4 cm-thicked PMMA slab used in conjunction with a RW3 polystyrene phantom (PTW, Freiburg, Germany) as shown in Fig. 2. This configuration allowed the acquisition of planar dose maps with a 2D array of ion chambers [11]. The TRILOGY phantom has been specifically built for this experiment. As depicted in Fig. 3 it consists of a plexiglass slab with three cavities where metallic cylinders can be inserted. In this configuration the point of measurements was closer to the cylinder (2.4 cm from the cylinder central axis versus 3.5 cm of configuration (1)). TRILOGY can be also used to reproduce both the conditions of single and bilateral hip prostheses implant. In both experiments the SAD was set to 100 cm, so the centre of cylinders is at a depth of 5 cm and reproduces a pelvis irradiation with a lateral beam. Clinical (10 10) cm2 6 MV and 18 MV photon beams were provided by a Siemens ONCOR Impression plus linear accelerator and incorporating an Optifocus 82 leaves MLC. 2.2. Monte Carlo simulations The MC code used in this investigation was BEAMnrc [12]. BEAMnrc was initially used to build a virtual model of the medical linear accelerator Siemens ONCOR Impression plus. An accurate Table 1 Prosthetic implant materials and corresponding physical density (Table reproduced with the permission from Spezi et al. [10]) Materials
Trade names
Standard
Density (g/cm3)
Titanium alloy
Ti6Al4V Ti6Al7Nb Vitallium Orthinox Stainless steel UHMWPE
ISO 5832/3 ISO5832/11 ISO 5832/9 ISO 5832/9 ISO 5832/1 ISO 5834/2
4.43 4.52 8.30 8.00 8.00 0.95
CoCr alloy Steel Polyethilene
Fig. 1. Experimental set-up (1): scheme of transverse and sagittal views (drawings not in scale). The cylinder rests on a specifically built support inside the water tank.
Fig. 2. Experimental set-up (2): scheme of the transverse view of TRILOGY (PMMA slab with cylindrical inserts) used in conjunction with a RW3 polystyrene phantom and a 2D array of ion chambers (drawings not in scale).
MC phase space (phsp) representation for both 6 MV and 18 MV energies of the linac was commissioned for field size ranging from (4 4) cm2 to (20 20) cm2. Phase space files have been generated in a plane located at 40 cm from the electron source. The density of the particles stored in the phsp files was about 10,000/mm2. MC models were validated through the comparison with experimental measurements carried out in a water phantom with a semiflex ion chamber (type 31010, PTW, Freiburg, Germany). Fig. 4(a) and (b) show the percentage depth dose (PDD) calculated with the DOSXYZnrc user code and PDD obtained from experimental measurements for 6 MV and 18 MV fields of several field sizes. Similarly, Fig. 5(a) and (b) shows the comparison between MC simulated and measured dose profiles at several depths for a 6 MV and
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
5003
Fig. 3. The TRILOGY phantom with cylinders of the different materials used in this experiment. Cylinders are 15 cm long and have a diameter of 3 cm.
Fig. 5. Profiles comparison between MC simulations and measured dose profiles for several depths for a (10 10) cm2 field size and for energies of 6 MV (a) and 18 MV (b).
18 MV (10 10) cm2 irradiating field. For both PDD and profiles, dose values have been independently normalized to dmax for the (10 10) cm2 field. Since the statistical uncertainty associated with the MC calculation was between 1% and 2% the size of the markers is comparable with the dimension of the error bar which is therefore not reported. A good agreement was found between MC calculations and experimental data: the difference was less than 2%. The experimental conditions described in Section 2.1 were reproduced within the MC simulation environment through the development of synthetic phantoms with a (2 2 5) mm3 grid resolution. Fig. 6(a) depicts the set-up condition for the irradiation of cylinders immersed in water phantom (cf. Fig. 1). Fig. 6(b) shows the set-up condition for the irradiation of the TRILOGY phantom (cf. Fig. 2). In this case the phantom was built from a CT scan of TRILOGY with plexiglass cylinders inserted. We did not CT scan the metallic cylinders in order to avoid image artefacts [4]. The presence of the different material inserts was instead reproduced by assigning the appropriate material composition and density to the correct 3D region of interest. Material composition and densities required for the MC modelling and are given in Tables 1 and 2. MC simulations were carried out in the same identical conditions for both 6 MV and 18 MV photon beams. 2.3. TPS calculations Fig. 4. PDD comparison between MC simulations and experimental measurements for (4 4), (7 7), (10 10) and (20 20) cm2 field sizes, for energies of 6 MV (a) and 18 MV (b).
The convolution algorithm used in the Pinnacle3 treatment planning system computes the dose distribution by modelling
5004
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
3. Results 3.1. MC simulations The comparison between the profiles measured with ion chamber and calculated with MC shows a very good agreement. Profiles were taken at a depth of 8.5 cm from the water surface (i.e. 3.5 cm below the isocentre and central axis of the cylinder). This limit was determined by the presence of the arm support, which the ion chamber was attached to inside the water tank. As far as 2D ARRAY measurements are concerned, the calculated profiles fit well (within 2% global difference) experimental measurements for all the considered materials and for both beam energies. Profiles were taken 2.4 cm below the isocentre. Also in this case the size of the markers is comparable with the dimension of the error bar which is therefore not reported. Some relevant comparisons between dose profiles measured and calculated with MC are depicted in our previous work [10]. 3.2. TPS calculations
Fig. 6. Set-up condition for the irradiation of cylinders in water phantom (sagittal view, cf. Fig. 1(a)). Set-up condition for the irradiation of the TRILOGY phantom (cf. Fig. 2(b)).
the incident energy fluence, which is projected through the density representation of a patient to compute TERMA (Total Energy Released per unit Mass). A superposition of the TERMA with a polyenergetic energy deposition kernel is used to compute the dose. Superposition is performed using collapsed-cones (CC), which refer to the modelling of a cone in space using a single ray corresponding to the central axis of the cone [13]. As for the MC simulations discussed in Section 2.2, we used synthetic phantoms to reproduce the experimental irradiation condition in the Pinnacle3 TPS (cf. Figs. 1 and 2). We manually assigned to the different inserts the appropriate densities given in Table 1. The standard CT conversion table of the TPS was extended to account for the increased density of the materials used for dose calculation. For both configurations we gave the same number of monitor units. PDD and dose profiles have been calculated with the Pinnacle3 CC algorithm commissioned and clinically used in our centre. The voxel size of TPS phantom dose calculation grid was (4 4 4) mm3. This is the standard dose calculation grid in Pinnacle3. This is different from the resolution of the MC dose calculation grid. Mora et al. [14] showed that the voxel size effect can be relevant in the case of MC dose calculations carried out with dose grids of different sizes. However, internal studies pointed out that the dose distribution calculated by TPS does not differ significantly for voxel sizes ranging between 1 mm3 and 4 mm3. The resolution of the TPS calculation grid was changed to 1 mm3 voxel size when we focused on the dose distributions near the interface tissue/prosthesis.
The comparison between ion chamber measurements and TPS water profiles for 6 MV at 8.5 cm depth is given in Fig. 7. Profiles include the titanium alloys, the steel alloys, Co–Cr and UHMWPE. Experimental data and TPS calculations were independently normalized to the value recorded/calculated at the centre of the open field at a depth of 8.5 cm. Overall, we find that TPS calculations match very well the measurements carried out in water. The highest difference, reported as expected for the cylinder with the highest density (Co–Cr), is less than 3%. The comparison between 2D array-measurements and TPS profiles for 18 MV shows a very good agreement as reported in Fig. 8. Profiles were taken at 2.4 cm below the isocentre, and an independent normalization to the central axis dose was carried out for the irradiation of a homogeneous RW3 polystyrene phantom. Results show that for all materials Pinnacle3 calculations fit very well experimental measurements, for both the energies and the experimental configurations. Furthermore, we evaluated how the CC algorithm implemented in Pinnacle3 models the dose distribution at the water–metal interface by comparing calculated PDD with MC simulations. We compared the central axis PDD for the set-up condition shown in Fig. 1. Dose distributions were calculated with both TPS and MC engines for a (10 10) cm2 field. Dose values have been independently normalized to a depth of 3.3 cm. The energy of the irradiating beam was set to 18 MV, which is the modality used in our centre for pelvic treatments. We evaluated the differences between the dose distributions at water–metal interface for titanium, steel and CoCr. PDDs and relative percentage difference, defined as (DTPS DMC)/DMC*100, for the three materials, are given in Fig. 9. The size of the MC calculations markers is comparable with the dimension of the error bar which is therefore not reported. The results show that interface effects, characterised by a loss of electronic equilibrium, are not correctly modelled by the CC algorithm: within 5 mm near the water–metal interface Pinnacle3 underestimates the dose by a maximum of 13% for titanium and steel and 16% for CoCr. The dose underestimation near the interfaces increases with the increasing of the density of the inhomogeneity. 4. Discussion and conclusions In this work we evaluated the capabilities of the dose calculation algorithm CC convolution implemented in the commercial TPS Pinnacle3 in presence of high-density materials. This investiga-
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
5005
Fig. 7. Profile comparison for measurements and TPS calculations of the 6 MV photon beam (cf. Fig. 1).
Fig. 8. Profile comparison for measurements and TPS calculations of the 18 MV photon beam (cf. Fig. 2).
tion is essential for the validation of the calculated dose distribution for patients with hip prosthesis undergoing external beam radiotherapy.
We investigated six different materials commonly used to manufacture hip prostheses. These data were taken from a regional register of orthopaedic prosthetic implants.
5006
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
calculation of clinical dose distributions when the information regarding the type of implanted material is available and when artifacts do not completely degrade image quality. According to our experience implants with a density higher than 4.5 g/cm3 generate relevant artefacts. The intensity and the extent of these artefacts increase for higher material density. Although there have been studies investigating, with some degree of success, the removal of metal artefacts [15] this technology does not seems to be readily available yet. On the other hand, the results of this investigation, obtained in ideal conditions, provide important information that can be used when evaluating dose distribution for a plan including prosthetic implants that do no generate high intensity artefacts. A study of the performance of both TPS and MC calculations with different level of image artefacts, in both ideal and clinical conditions, could produce useful data to be used in treatment planning. Finally, the dose validation method used in this investigation could be successfully extended to other clinical sites and used in the clinical practice. Acknowledgments We thank Adler Ortho s.r.l. (Milan, Italy), Hit Medica s.r.l. (Rimini, Italy) and Marle s.a. (Odival, France) for providing the prosthetic implant materials used in this work. We are also grateful to the Register of the Orthopaedic Prosthetic Implants (RIPO, Istituti Ortopedici Rizzoli, Bologna, Italy) for the support with the database of implanted materials. Ing W. Leardini (Istituti Ortopedici Rizzoli, Bologna, Italy) and Mr A. Angelini (TOMA s.n.c., Pesaro, Italy) are acknowledged for their help with the manufacture of high-density cylinders. References
Fig. 9. PDD comparison between MC and TPS calculations for 18 MV photon beam (cf. Fig. 1) in presence of a cylinder of titanium (a), steel (b) and CoCr (c).
For a 0.95–8.30 g/cm3 range of density, we have shown that, once assigned the correct density to the implant, Pinnacle3 succeeds in calculating the dose shadowing introduced by the prostheses with a maximum error of 3% within 0.5 cm from the high-Z cavity. This result has been achieved in two different experimental configurations. We also evaluated the capability of collapsed-cone convolution algorithm in calculating dose distribution near high-Z interface by comparison with MC calculations. Close to the water–metal interface Pinnacle3 underestimates the dose respect to MC calculations and this difference increases with the density of the prosthetic material. However, the problems associated with dose differences at the interfaces are regarded of less importance in the clinical practice when dealing with pelvic irradiation in presence of hip prosthesis. This is because the hip prosthesis is seldom irradiated by a primary beam and because the extent of the dose peaks are of the order of magnitude of the range of secondary electrons. However, this may not always be possible and the accurate dose calculation at material interfaces can indeed help minimizing the dose received by the tissue surrounding the implant. The MC characterization of the materials of hip implants is an important clinical resource: the analysis of the dosimetric properties of these materials is a fundamental condition for the accurate
[1] C. Reft, R. Alecu, I.J. Das, B.J. Gerbi, P. Keall, E. Lief, B.J. Mijnheer, N. Papanikolaou, C. Sibata, J. Van Dyk, Dosimetric considerations for patients with hip prostheses undergoing pelvic irradiation. Report of the AAPM Radiation Therapy Committee Task Group 63, Med. Phys. 30 (2003), p. 1162. [2] R. Roberts, How accurate is a CT-based dose calculation on a pencil beam TPS for a patient with a metallic prosthesis, Phys. Med. Biol. 46 (2001) N227. [3] G. Ding, C. Yu, A study on beams passing through hip prosthesis for pelvic radiation treatment, IJROBP 51 (2001) 1167. [4] S. Lin, T. Chu, J. Lin, M. Liu, The effect of a metal hip prosthesis on the radiation dose in therapeutic photon beam irradiations, Appl. Radiat. Isot. 57 (2002) 17. [5] E. Wieslander, T. Knöös, Dose perturbation in the presence of metallic implants: treatment planning system versus Monte Carlo simulation, Phys. Med. Biol. 48 (2003) 3295. [6] W. Laub, F. Nüsslin, Monte Carlo dose calculations in the treatment of a pelvis with implant and comparison with pencil-beam calculations, Med. Dos. 28 (2003) 229. [7] P.J. Keall, J.V. Siebers, R. Jeraj, R. Mohan, Radiotherapy dose calculations in the presence of hip prostheses, Med. Dos. 28 (2003) 107. [8] E. Buffard, R. Gschwind, L. Makovicka, E. Martin, C. Meunier, C. David, Study of the impact of artificial articulations on the dose distribution under medical irradiation, Nucl. Instr. and Meth. B 229 (2005) 78. [9] E. Buffard, R. Gschwind, L. Makovicka, C. David, Monte Carlo calculations of the impact of a hip prosthesis on the dose distribution, Nucl. Instr. and Meth. B 251 (2006) 9. [10] E. Spezi, F. Palleri, A.L. Angelini, A. Ferri, F. Baruffaldi, Characterization of materials for prosthetic implants using the BEAMnrc Monte Carlo code, First European Workshop on Monte Carlo Treatment Planning, J. Phys. Conf. Series 74 (2007) 012016. [11] E. Spezi, A.L. Angelini, F. Romani, A. Ferri, Characterization of a 2D ion chamber array for the verification of radiotherapy treatments, Phys. Med. Biol. 50 (2005) 3361. [12] D.W.O. Rogers, B.A. Faddegon, G.X. Ding, C.-M. Ma, J. Wei, T.R. Mackie, BEAM: A Monte Carlo code to simulate radiotherapy treatment units, Med. Phys. 22 (1995) 503. [13] Pinnacle3 Physics REFERENCE GUIDE. [14] G. Mora, T. Pawlicki, A. Maio, C.-M. Ma, Effect of voxel size on Monte Carlo dose calculations for radiotherapy treatment planning, in: Proceedings of the Monte Carlo 2000 Conference Advanced Monte Carlo for Radiation Physics, Particle Transport Simulations and Applications, Springer-Verlag, Berlin, 2001, p. 549. [15] M. Yazdia, L. Gingras, L. Beaulieu, An adaptive approach to metal artifact reduction in helical computed tomography for radiation therapy treatment planning: experimental and clinical studies, IJROBP 62 (2005) 1224.
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Monte Carlo characterization of materials for prosthetic implants and dosimetric validation of Pinnacle3 TPS q Francesca Palleri a,*, Fabio Baruffaldi b, Anna Lisa Angelini a, Andrea Ferri a,1, Emiliano Spezi a,2 a b
Servizio di Fisica Sanitaria, Policlinico S.Orsola-Malpighi, Via Massarenti 9, 40138 Bologna, Italy Laboratorio di Tecnologia Medica, Istituti Ortopedici Rizzoli, via di Barbiano 1/10, 40136 Bologna, Italy
a r t i c l e
i n f o
Article history: Received 11 June 2008 Received in revised form 26 August 2008 Available online 9 September 2008 Keywords: Hip prosthesis Dose calculation Treatment planning system BEAMnrc Monte Carlo code
a b s t r a c t In external beam radiotherapy the calculation of dose distribution for patients with hip prostheses is critical. Metallic implants not only degrade the image quality but also perturb the dose distribution. Conventional treatment planning systems do not accurately account for high-Z prosthetic implants heterogeneities, especially at interfaces. The materials studied in this work have been chosen on the basis of a statistical investigation on the hip prostheses implanted in 70 medical centres. The first aim of this study is a systematic characterization of materials used for hip prostheses, and it has been provided by BEAMnrc Monte Carlo code. The second aim is to evaluate the capabilities of a specific treatment planning system, Pinnacle3, when dealing with dose calculations in presence of metals, also close to the regions of high-Z gradients. In both cases it has been carried out an accurate comparison versus experimental measurements for two clinical photon beam energies (6 MV and 18 MV) and for two experimental sets-up: metallic cylinders inserted in a water phantom and in a specifically built PMMA slab. Our results show an agreement within 2% between experiments and MC simulations. TPS calculations agree with experiments within 3%. Ó 2008 Elsevier B.V. All rights reserved.
1. Introduction The age of population is growing and more patients with hip prostheses are undergoing external beam radiotherapy for pelvic malignancies. Dose planning for patients with metallic implants is challenging because high atomic number prostheses influence heavily the dose distribution prescribed to the target and to the surrounding tissues. The dosimetric aspects of external radiotherapy of patients with hip prostheses have been reviewed and special beam arrangements, in which prosthesis must preferably be avoided, have been advised [1]. Nevertheless, avoiding the hip prosthesis at the time of planning it is not always possible, because it may compromise the dose sparing of the rectum and the bladder. In external beam radiotherapy a detailed knowledge of the dose perturbation introduced by a high-Z prosthetic implant is a very q Part of this work has been presented to the First European Workshop on Monte Carlo Treatment Planning (Ghent 2006). * Corresponding author. Tel.: +39 0516363133; fax: +39 0516363571. E-mail addresses: [email protected], [email protected] (F. Palleri). 1 Present address: Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori (I.R.S.T.), Via Maroncelli 40, 47014 Meldola (FC), Italy. 2 Present address: Department of Medical Physics, Velindre Cancer Centre, CF142TL Cardiff (UK).
0168-583X/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2008.08.013
important treatment planning resource for the correct calculation of the dose distribution delivered to patients. Therefore, knowledge of capabilities and limitations of the treatment planning system in these situations is required. Roberts [2] pointed out the importance of evaluating the accuracy of the dose calculation carried out with a conventional treatment planning system (TPS) by recording dose profiles under a metallic rod immersed in water and by comparing them with calculated profiles. Monte Carlo (MC) methods allow dosimetric characteristics of high-density materials to be evaluated. Many studies deal with MC dose calculation in presence of hip prostheses. In the most recent works there are comparisons between dose calculations carried out with MC methods and with the most common algorithms implemented in TPS [3–7]. Nevertheless these works did not validate MC simulations or TPS calculations with experimental measurements. Recently Buffard et al. [8,9] investigated MC dose calculation for the irradiation of metal alloys; in particular they studied the interface effects, which are local dose perturbations in presence of high-Z gradients. However, these works did not study the effect of these materials at more common photon energies used in radiotherapy treatment planning. In this work we evaluated the dosimetric characteristics of some of the most common materials used for prosthetic implants. We carried out an accurate research to study the characteristics of the prosthetic materials actually implanted in the centres included
5002
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
in this study. The validation of our MC models versus experimental data for clinical 6 MV and 18 MV photon beams was carried out in a earlier study [10]. However, in that study we did not validate TPS calculations against measurements or MC calculations. In the present work we evaluated the collapsed-cones convolution (CC) algorithm employed in Pinnacle3 TPS (Philips, The Netherlands) in presence of high-density materials through the comparison with experimental data, and we focused on the way CC algorithm accounts for interface effects. 2. Materials and methods
Table 2 Chemical compositions of the most common prosthetic materials Titanium alloys
CoCr alloys
Steel alloys
Element
Percent
Element
Percent
Element
Percent
Ti Al V Fe O C N H
Balanced 5.5–6.8 3.5–4.5 0.3 max 0.2 max 0.06 max 0.05 max 0.015 max
Co Cr Mo Ni Fe Mn Si C N
Balanced 26.5–30.0 4.5–7.0 0.1 max 0.1 max 0.1 max 0.1 max 0.1 max 0.25 max
Fe Cr Ni Mn Mo Si N Cu C
Balanced 17.0–19.0 13.0–16.0 2.0 max 2.3–4.2 1.0 max 0.1–0.2 max 0.5 0.03
2.1. Experimental measurements The materials studied in this work have been chosen on the basis of the regional register of orthopaedic prosthetic implants (http://ripo.cineca.it). The 80% of stems and 70% of acetabular cups of the hip prostheses implanted between 2000 and 2004 were manufactured with six materials: two types of titanium alloy, two types of steel alloy, CoCr alloy and ultra high molecular weight polyethylene (UHMWPE). The full list of materials is given in Table 1, with density, international standard code and international trade name. The chemical composition, as provided by the manufacturers, is given in Table 2. We carried out a MC model for these materials, which have been produced in cylinders of 15 cm length and 3 cm diameter for this investigation. For our experiment we used two configurations: (1) the cylinders were immersed in a water phantom (MP3/PTW, Freiburg, Germany). Dose profiles at different depths under the cylinders were acquired with a semiflex ion chamber (type 31010, PTW, Freiburg, Germany) with a sensitive volume of 0.125 cm3. The experimental set-up is shown in Fig. 1; (2) the cylinders were inserted in TRILOGY, a 4 cm-thicked PMMA slab used in conjunction with a RW3 polystyrene phantom (PTW, Freiburg, Germany) as shown in Fig. 2. This configuration allowed the acquisition of planar dose maps with a 2D array of ion chambers [11]. The TRILOGY phantom has been specifically built for this experiment. As depicted in Fig. 3 it consists of a plexiglass slab with three cavities where metallic cylinders can be inserted. In this configuration the point of measurements was closer to the cylinder (2.4 cm from the cylinder central axis versus 3.5 cm of configuration (1)). TRILOGY can be also used to reproduce both the conditions of single and bilateral hip prostheses implant. In both experiments the SAD was set to 100 cm, so the centre of cylinders is at a depth of 5 cm and reproduces a pelvis irradiation with a lateral beam. Clinical (10 10) cm2 6 MV and 18 MV photon beams were provided by a Siemens ONCOR Impression plus linear accelerator and incorporating an Optifocus 82 leaves MLC. 2.2. Monte Carlo simulations The MC code used in this investigation was BEAMnrc [12]. BEAMnrc was initially used to build a virtual model of the medical linear accelerator Siemens ONCOR Impression plus. An accurate Table 1 Prosthetic implant materials and corresponding physical density (Table reproduced with the permission from Spezi et al. [10]) Materials
Trade names
Standard
Density (g/cm3)
Titanium alloy
Ti6Al4V Ti6Al7Nb Vitallium Orthinox Stainless steel UHMWPE
ISO 5832/3 ISO5832/11 ISO 5832/9 ISO 5832/9 ISO 5832/1 ISO 5834/2
4.43 4.52 8.30 8.00 8.00 0.95
CoCr alloy Steel Polyethilene
Fig. 1. Experimental set-up (1): scheme of transverse and sagittal views (drawings not in scale). The cylinder rests on a specifically built support inside the water tank.
Fig. 2. Experimental set-up (2): scheme of the transverse view of TRILOGY (PMMA slab with cylindrical inserts) used in conjunction with a RW3 polystyrene phantom and a 2D array of ion chambers (drawings not in scale).
MC phase space (phsp) representation for both 6 MV and 18 MV energies of the linac was commissioned for field size ranging from (4 4) cm2 to (20 20) cm2. Phase space files have been generated in a plane located at 40 cm from the electron source. The density of the particles stored in the phsp files was about 10,000/mm2. MC models were validated through the comparison with experimental measurements carried out in a water phantom with a semiflex ion chamber (type 31010, PTW, Freiburg, Germany). Fig. 4(a) and (b) show the percentage depth dose (PDD) calculated with the DOSXYZnrc user code and PDD obtained from experimental measurements for 6 MV and 18 MV fields of several field sizes. Similarly, Fig. 5(a) and (b) shows the comparison between MC simulated and measured dose profiles at several depths for a 6 MV and
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
5003
Fig. 3. The TRILOGY phantom with cylinders of the different materials used in this experiment. Cylinders are 15 cm long and have a diameter of 3 cm.
Fig. 5. Profiles comparison between MC simulations and measured dose profiles for several depths for a (10 10) cm2 field size and for energies of 6 MV (a) and 18 MV (b).
18 MV (10 10) cm2 irradiating field. For both PDD and profiles, dose values have been independently normalized to dmax for the (10 10) cm2 field. Since the statistical uncertainty associated with the MC calculation was between 1% and 2% the size of the markers is comparable with the dimension of the error bar which is therefore not reported. A good agreement was found between MC calculations and experimental data: the difference was less than 2%. The experimental conditions described in Section 2.1 were reproduced within the MC simulation environment through the development of synthetic phantoms with a (2 2 5) mm3 grid resolution. Fig. 6(a) depicts the set-up condition for the irradiation of cylinders immersed in water phantom (cf. Fig. 1). Fig. 6(b) shows the set-up condition for the irradiation of the TRILOGY phantom (cf. Fig. 2). In this case the phantom was built from a CT scan of TRILOGY with plexiglass cylinders inserted. We did not CT scan the metallic cylinders in order to avoid image artefacts [4]. The presence of the different material inserts was instead reproduced by assigning the appropriate material composition and density to the correct 3D region of interest. Material composition and densities required for the MC modelling and are given in Tables 1 and 2. MC simulations were carried out in the same identical conditions for both 6 MV and 18 MV photon beams. 2.3. TPS calculations Fig. 4. PDD comparison between MC simulations and experimental measurements for (4 4), (7 7), (10 10) and (20 20) cm2 field sizes, for energies of 6 MV (a) and 18 MV (b).
The convolution algorithm used in the Pinnacle3 treatment planning system computes the dose distribution by modelling
5004
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
3. Results 3.1. MC simulations The comparison between the profiles measured with ion chamber and calculated with MC shows a very good agreement. Profiles were taken at a depth of 8.5 cm from the water surface (i.e. 3.5 cm below the isocentre and central axis of the cylinder). This limit was determined by the presence of the arm support, which the ion chamber was attached to inside the water tank. As far as 2D ARRAY measurements are concerned, the calculated profiles fit well (within 2% global difference) experimental measurements for all the considered materials and for both beam energies. Profiles were taken 2.4 cm below the isocentre. Also in this case the size of the markers is comparable with the dimension of the error bar which is therefore not reported. Some relevant comparisons between dose profiles measured and calculated with MC are depicted in our previous work [10]. 3.2. TPS calculations
Fig. 6. Set-up condition for the irradiation of cylinders in water phantom (sagittal view, cf. Fig. 1(a)). Set-up condition for the irradiation of the TRILOGY phantom (cf. Fig. 2(b)).
the incident energy fluence, which is projected through the density representation of a patient to compute TERMA (Total Energy Released per unit Mass). A superposition of the TERMA with a polyenergetic energy deposition kernel is used to compute the dose. Superposition is performed using collapsed-cones (CC), which refer to the modelling of a cone in space using a single ray corresponding to the central axis of the cone [13]. As for the MC simulations discussed in Section 2.2, we used synthetic phantoms to reproduce the experimental irradiation condition in the Pinnacle3 TPS (cf. Figs. 1 and 2). We manually assigned to the different inserts the appropriate densities given in Table 1. The standard CT conversion table of the TPS was extended to account for the increased density of the materials used for dose calculation. For both configurations we gave the same number of monitor units. PDD and dose profiles have been calculated with the Pinnacle3 CC algorithm commissioned and clinically used in our centre. The voxel size of TPS phantom dose calculation grid was (4 4 4) mm3. This is the standard dose calculation grid in Pinnacle3. This is different from the resolution of the MC dose calculation grid. Mora et al. [14] showed that the voxel size effect can be relevant in the case of MC dose calculations carried out with dose grids of different sizes. However, internal studies pointed out that the dose distribution calculated by TPS does not differ significantly for voxel sizes ranging between 1 mm3 and 4 mm3. The resolution of the TPS calculation grid was changed to 1 mm3 voxel size when we focused on the dose distributions near the interface tissue/prosthesis.
The comparison between ion chamber measurements and TPS water profiles for 6 MV at 8.5 cm depth is given in Fig. 7. Profiles include the titanium alloys, the steel alloys, Co–Cr and UHMWPE. Experimental data and TPS calculations were independently normalized to the value recorded/calculated at the centre of the open field at a depth of 8.5 cm. Overall, we find that TPS calculations match very well the measurements carried out in water. The highest difference, reported as expected for the cylinder with the highest density (Co–Cr), is less than 3%. The comparison between 2D array-measurements and TPS profiles for 18 MV shows a very good agreement as reported in Fig. 8. Profiles were taken at 2.4 cm below the isocentre, and an independent normalization to the central axis dose was carried out for the irradiation of a homogeneous RW3 polystyrene phantom. Results show that for all materials Pinnacle3 calculations fit very well experimental measurements, for both the energies and the experimental configurations. Furthermore, we evaluated how the CC algorithm implemented in Pinnacle3 models the dose distribution at the water–metal interface by comparing calculated PDD with MC simulations. We compared the central axis PDD for the set-up condition shown in Fig. 1. Dose distributions were calculated with both TPS and MC engines for a (10 10) cm2 field. Dose values have been independently normalized to a depth of 3.3 cm. The energy of the irradiating beam was set to 18 MV, which is the modality used in our centre for pelvic treatments. We evaluated the differences between the dose distributions at water–metal interface for titanium, steel and CoCr. PDDs and relative percentage difference, defined as (DTPS DMC)/DMC*100, for the three materials, are given in Fig. 9. The size of the MC calculations markers is comparable with the dimension of the error bar which is therefore not reported. The results show that interface effects, characterised by a loss of electronic equilibrium, are not correctly modelled by the CC algorithm: within 5 mm near the water–metal interface Pinnacle3 underestimates the dose by a maximum of 13% for titanium and steel and 16% for CoCr. The dose underestimation near the interfaces increases with the increasing of the density of the inhomogeneity. 4. Discussion and conclusions In this work we evaluated the capabilities of the dose calculation algorithm CC convolution implemented in the commercial TPS Pinnacle3 in presence of high-density materials. This investiga-
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
5005
Fig. 7. Profile comparison for measurements and TPS calculations of the 6 MV photon beam (cf. Fig. 1).
Fig. 8. Profile comparison for measurements and TPS calculations of the 18 MV photon beam (cf. Fig. 2).
tion is essential for the validation of the calculated dose distribution for patients with hip prosthesis undergoing external beam radiotherapy.
We investigated six different materials commonly used to manufacture hip prostheses. These data were taken from a regional register of orthopaedic prosthetic implants.
5006
F. Palleri et al. / Nuclear Instruments and Methods in Physics Research B 266 (2008) 5001–5006
calculation of clinical dose distributions when the information regarding the type of implanted material is available and when artifacts do not completely degrade image quality. According to our experience implants with a density higher than 4.5 g/cm3 generate relevant artefacts. The intensity and the extent of these artefacts increase for higher material density. Although there have been studies investigating, with some degree of success, the removal of metal artefacts [15] this technology does not seems to be readily available yet. On the other hand, the results of this investigation, obtained in ideal conditions, provide important information that can be used when evaluating dose distribution for a plan including prosthetic implants that do no generate high intensity artefacts. A study of the performance of both TPS and MC calculations with different level of image artefacts, in both ideal and clinical conditions, could produce useful data to be used in treatment planning. Finally, the dose validation method used in this investigation could be successfully extended to other clinical sites and used in the clinical practice. Acknowledgments We thank Adler Ortho s.r.l. (Milan, Italy), Hit Medica s.r.l. (Rimini, Italy) and Marle s.a. (Odival, France) for providing the prosthetic implant materials used in this work. We are also grateful to the Register of the Orthopaedic Prosthetic Implants (RIPO, Istituti Ortopedici Rizzoli, Bologna, Italy) for the support with the database of implanted materials. Ing W. Leardini (Istituti Ortopedici Rizzoli, Bologna, Italy) and Mr A. Angelini (TOMA s.n.c., Pesaro, Italy) are acknowledged for their help with the manufacture of high-density cylinders. References
Fig. 9. PDD comparison between MC and TPS calculations for 18 MV photon beam (cf. Fig. 1) in presence of a cylinder of titanium (a), steel (b) and CoCr (c).
For a 0.95–8.30 g/cm3 range of density, we have shown that, once assigned the correct density to the implant, Pinnacle3 succeeds in calculating the dose shadowing introduced by the prostheses with a maximum error of 3% within 0.5 cm from the high-Z cavity. This result has been achieved in two different experimental configurations. We also evaluated the capability of collapsed-cone convolution algorithm in calculating dose distribution near high-Z interface by comparison with MC calculations. Close to the water–metal interface Pinnacle3 underestimates the dose respect to MC calculations and this difference increases with the density of the prosthetic material. However, the problems associated with dose differences at the interfaces are regarded of less importance in the clinical practice when dealing with pelvic irradiation in presence of hip prosthesis. This is because the hip prosthesis is seldom irradiated by a primary beam and because the extent of the dose peaks are of the order of magnitude of the range of secondary electrons. However, this may not always be possible and the accurate dose calculation at material interfaces can indeed help minimizing the dose received by the tissue surrounding the implant. The MC characterization of the materials of hip implants is an important clinical resource: the analysis of the dosimetric properties of these materials is a fundamental condition for the accurate
[1] C. Reft, R. Alecu, I.J. Das, B.J. Gerbi, P. Keall, E. Lief, B.J. Mijnheer, N. Papanikolaou, C. Sibata, J. Van Dyk, Dosimetric considerations for patients with hip prostheses undergoing pelvic irradiation. Report of the AAPM Radiation Therapy Committee Task Group 63, Med. Phys. 30 (2003), p. 1162. [2] R. Roberts, How accurate is a CT-based dose calculation on a pencil beam TPS for a patient with a metallic prosthesis, Phys. Med. Biol. 46 (2001) N227. [3] G. Ding, C. Yu, A study on beams passing through hip prosthesis for pelvic radiation treatment, IJROBP 51 (2001) 1167. [4] S. Lin, T. Chu, J. Lin, M. Liu, The effect of a metal hip prosthesis on the radiation dose in therapeutic photon beam irradiations, Appl. Radiat. Isot. 57 (2002) 17. [5] E. Wieslander, T. Knöös, Dose perturbation in the presence of metallic implants: treatment planning system versus Monte Carlo simulation, Phys. Med. Biol. 48 (2003) 3295. [6] W. Laub, F. Nüsslin, Monte Carlo dose calculations in the treatment of a pelvis with implant and comparison with pencil-beam calculations, Med. Dos. 28 (2003) 229. [7] P.J. Keall, J.V. Siebers, R. Jeraj, R. Mohan, Radiotherapy dose calculations in the presence of hip prostheses, Med. Dos. 28 (2003) 107. [8] E. Buffard, R. Gschwind, L. Makovicka, E. Martin, C. Meunier, C. David, Study of the impact of artificial articulations on the dose distribution under medical irradiation, Nucl. Instr. and Meth. B 229 (2005) 78. [9] E. Buffard, R. Gschwind, L. Makovicka, C. David, Monte Carlo calculations of the impact of a hip prosthesis on the dose distribution, Nucl. Instr. and Meth. B 251 (2006) 9. [10] E. Spezi, F. Palleri, A.L. Angelini, A. Ferri, F. Baruffaldi, Characterization of materials for prosthetic implants using the BEAMnrc Monte Carlo code, First European Workshop on Monte Carlo Treatment Planning, J. Phys. Conf. Series 74 (2007) 012016. [11] E. Spezi, A.L. Angelini, F. Romani, A. Ferri, Characterization of a 2D ion chamber array for the verification of radiotherapy treatments, Phys. Med. Biol. 50 (2005) 3361. [12] D.W.O. Rogers, B.A. Faddegon, G.X. Ding, C.-M. Ma, J. Wei, T.R. Mackie, BEAM: A Monte Carlo code to simulate radiotherapy treatment units, Med. Phys. 22 (1995) 503. [13] Pinnacle3 Physics REFERENCE GUIDE. [14] G. Mora, T. Pawlicki, A. Maio, C.-M. Ma, Effect of voxel size on Monte Carlo dose calculations for radiotherapy treatment planning, in: Proceedings of the Monte Carlo 2000 Conference Advanced Monte Carlo for Radiation Physics, Particle Transport Simulations and Applications, Springer-Verlag, Berlin, 2001, p. 549. [15] M. Yazdia, L. Gingras, L. Beaulieu, An adaptive approach to metal artifact reduction in helical computed tomography for radiation therapy treatment planning: experimental and clinical studies, IJROBP 62 (2005) 1224.