- Email: [email protected]
Comparison of two dose calculation methods applied to extracranial stereotactic radiotherapy treatment planning
Radiotherapy and Oncology 77 (2005) 96–98 www.thegreenjournal.com
Extracranial stereotactic RT
Comparison of two dose calculation methods applied to extracranial stereotactic radiotherapy treatment planning Anders Traberg Hansena,*, Jørgen B. Petersena, Morten Høyerb, Jens Juul Christensena a
Department of Medical Physics, and bDepartment of Oncology, Aarhus University Hospital, Aarhus, Denmark
Abstract Background and purpose: Lung tissue is a special challenge for a dose calculation algorithm, especially in the case of extracranial stereotactic radiotherapy (ESRT) due to small field sizes in combination with large variations in tissue density. The present study investigates the choice of dose calculation algorithm for 18 patients with a single lung tumor and 8 patients with a single liver tumor. The dose calculation is performed with both the pencil beam convolution algorithm and the collapsed cone convolution algorithm with the same number of monitor units in both cases. In addition, the dose calculation with the collapsed cone convolution algorithm is also performed with modified field sizes in order to match the Planning Target Volume (PTV) peripheral dose of the pencil beam based treatment. Results: For liver tumors, the mean Clinical Target Volume (CTV) dose calculated by the collapsed cone convolution algorithm and the pencil beam convolution algorithm is almost identical. For lung tumors, the mean CTV dose determined by the collapsed cone convolution algorithm differs up to 20%. Plans obtained by the two algorithms have field sizes which differ up to 8 mm for the same number of monitor units and minimum dose to the lung PTV. Conclusions: The choice of dose calculation algorithm can have a large influence on a treatment plan for ESRT of the lungs. q 2005 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 77 (2005) 96–98. Keywords: Extracraniel stereotactic radiotherapy; Treatment planning; Dose calculation algorithms; Collapsed cone
In recent years, several radiotherapy centers have introduced extracranial stereotactic radiotherapy (ESRT) as a new modality for treatment of tumors in the liver and in the lung [3,4,8,10]. SRT is characterized by the use of several small fields for the treatment of small volumes with a high dose [2,4]. This is a challenge for the treatment planning system (TPS), especially in the lung, where the scatter conditions are very different to those in water. In this study, the most commonly used calculation algorithm, the pencil beam convolution algorithm, is compared to the collapsed cone convolution algorithm, which is generally accepted to be more accurate [1,6,7,11].
Materials and methods The patients that this study is based on are immobilized by the Electa stereotactic body frame [4]. CT scans of the patients are carried out with a slice thickness of 5.0 mm and the subsequent dose planning and dose calculation is performed by the HELAX-TMS TPS version 6.1 B. The CTV is set equal to the delineated Gross Tumor Volume (GTV) and the PTV is obtained by expanding the
boundary of the CTV. The centrally located ICRU point [5] is usually coinciding with the treatment isocenter. The PTV is included in the 67% isodose, and the CTV is included in the 95% isodose (Lax originally prescribed a dose of 100% to the PTV periphery, corresponding to a dose of approximately 150% in the central part of the tumor, which is similar to prescription methods used in cranial stereotactic radiosurgery) [4]. The treatment plans consist of 4–8 photon fields with an energy of 6 MV in either a coplanar or a noncoplaner field arrangement. The field sizes used range from 4!4 to 10!10 cm and distance between the calculation points is 4.5 mm and the fields are shaped by a Siemens MLC with a leaf width of 1 cm at the isocenter. All patients are treated with plans calculated by the pencil beam convolution algorithm. For each patient, the treatment plan is recalculated using the collapsed cone convolution algorithm with the same number of monitor units as that of the pencil beam convolution algorithm. All beam related parameters are the same for the two calculation methods. During the initial planning of a patient, the MLC-leaves are automatically adapted to a contour obtained from an expansion of the projected PTV boundary,
0167-8140/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2005.04.018
A. Traberg Hansen et al. / Radiotherapy and Oncology 77 (2005) 96–98
Results and discussion Fig. 1 shows the ratios between the minimum PTV dose and the mean GTV dose calculated by the pencil beam or the collapsed cone convolution algorithms for lung tumors. It is observed that for some of the small tumors, the minimum doses to the PTV are up to 65% larger for the pencil beam convolution algorithm than for the collapse cone convolution algorithm. The mean dose to the GTV is best for reporting dose, since this quantity have a small dependence of the used algorithm and therefore suitable for this purpose. The same analysis for the 8 patients with liver tumors shows that the doses calculated by the two calculation algorithms are within 1% from each other. Fig. 2 shows the effect of the different margins used. It is observed that an additional margin around the PTV of up to 4 mm is needed for the same number of monitor units to obtain the same minimum dose to the PTV for the collapsed cone convolution algorithm as that of the pencil beam convolution algorithm. The collapsed cone convolution algorithm takes the range of the scattered secondary
PP-cal/CC-cal
1.8
1.6
PTV min
1.4
GTV mean
1.2
1
0.8 2
5
13
23 38 48 70 GTV volume (ccm)
91
190
Fig. 1. Ratio of minimum dose in the lung PTV calculated with the pencil beam convolution algorithm and the collapsed cone convolution algorithm for the 18 lung tumors (†), and the corresponding ratio for the mean dose in the GTV (;).
25 48 ccm 27 ccm
15 Relative deviation (%)
and subsequently adjusted to cover the PTV by the 67% isodose curve. To examine the effect of adding margins to the PTV, a systematic adaptation of MLC-leaves is used. The MLC-leaves of the initial treatment plan are therefore automatically readapted to the contour in such a way that the midpoints of the leaves are placed on the contour. This plan is evaluated by the pencil beam convolution algorithm. Additional plans are made in the same way except that the MLC-leaves are adapted to the contour of the projected PTV boundary with an extra margin of 0–4 mm. These plans are evaluated by the collapsed cone convolution algorithm and renormalised to the same number of monitor units as that of the plan used for treatment, and the minimum dose to the PTV is recorded. This procedure is performed for five selected patients with lung tumors.
97
141 ccm 2 ccm
5
5 ccm -5
-15
-25 0
1
2 3 Extra margin (mm)
4
Fig. 2. Difference in the minimum dose to the PTV between several collapsed cone calculations and a pencil beam calculation as a function of the extra margin from the projected PTV contour to the field edge for five lung tumors.
electrons into account [1]. This is especially important in the lung, where the range of secondary electrons is more than twice that of water. This difference in the handling of the scattered electrons has profound implications on the algorithms ability to describe the penumbra region. It is therefore generally accepted that the collapsed cone convolution algorithm is superior to the pencil beam convolution algorithm for the calculation of radiation dose in lung tissue [1,3,6,7,11]. The pencil beam convolution algorithm tends to overestimate the dose to a small lung tumor in comparison to the collapsed cone convolution algorithm, as shown in Fig. 1, because the loss of scattered particles is significant for a small tumor. For a dose calculation in the liver tissue, the two algorithms predict almost the same doses, since the liver is more equivalent to water. In ESRT, the dose delivered to the tumor volume is not necessarily homogeneous. It has therefore been suggested to focus on the minimum dose delivered to the PTV [10]. However, in the case of a lung tumor, the minimum dose is typically located at the surface of the PTV, where there is a steep dose gradient. This makes the minimum dose to the PTV very sensitive to the choice of dose calculation algorithm. It is also possible to quantify the differences between the calculation algorithms by keeping the field weights, field arrangement, and monitor units fixed. A plan with a specified minimum dose to the PTV based on the pencil beam convolution algorithm can be reproduced by the collapsed cone convolution algorithm by increasing the margins, and by using the same number of monitor units. Instead of claiming that the pencil beam convolution algorithm overestimates doses to lung tumors, it can as well be claimed the pencil beam convolution algorithm underestimates the margins needed to give a prescribed minimum dose to the PTV. This study reveals that the underestimation of margins in some cases exceeds 4 mm. Omitted by this study are other sources of uncertainty in the process of ESRT for instance the important issue of tumor movement [9,10].
98
Comparison of two dose calculation methods
In conclusion, we find that the pencil beam convolution algorithm tends to overestimate the dose to small lung tumors relative to the collapsed cone convolution algorithm, which is corresponding to an underestimation the margins needed to secure a sufficient coverage of the PTV. Therefore, by using the pencil beam convolution algorithm, one could under dose small lung tumors significantly. We therefore recommend that a collapsed cone calculation is used for treatment of lung tumors. In addition, dose prescription to the periphery of a lung tumor should be avoided.
*
Corresponding author. Anders Traberg Hansen, Tel.: 89492594; fax: 89492590. E-mail address: [email protected] Received 25 May 2004; received in revised form 5 April 2005; accepted 6 April 2005
References [1] Aspradakis MM, Morrison RH, Richmond ND, Steele A. Experimental verification of convolution/superposition photon dose calculations for radiotherapy treatment planning. Phys Med Biol 2003;48:2873–93. [2] Brahme A. Dosimetric precision requirements in radiation therapy. Acta Radiol Oncol 1984;23:379–91.
[3] Butson MJ, et al. Verification of lung dose in an anthropomorphic phantom calculated by the collapsed cone convolution method. Phys Med Biol 2000;45:N143–N9. [4] Ingmar Lax, Henric Blomgren, Ingemar Na ¨slund, Rut Svanstro ¨m. Stereotactic radiotherapy of malignancies in the Abdomen. Acta Oncologica 1994;33:677–83. [5] ICRU (International Commission on Radiological Units) Report 50. Prescribing, recording, and reporting photon beam therapy. Washington, DC: ICRU; 1993. [6] Lydon JM. Photon dose calculations in homogeneous media for a treatment planning system using a collapsed cone superposition convolution algorithm. Phys Med Biol 1998;43:1813–22. [7] Weber L, Nilsson P. Verification of Dose Calculations with a Clinical Treatment Planning system based on a Point Kernel dose engine. J Appl Clin Med Phys 2002;3(2):73–87. [8] Wulf J, et al. Stereotactic radiotherapy for primary lung cancer and pulmonary metastases: a noninvasive treatment approach in medically inoperable patients. Int J Radiat Oncol Biol Phys 2004;60:186–96. [9] Wulf J, et al. Impact of target reproducibility on tumor dose in stereotactic radiotherapy of targets in the lung and liver. Radiother Oncol 2003;66:141–50. [10] Wulf J, et al. Stereotactic radiotherapy of extracranial targets: CT-simulation and accuracy of treatment in the stereotactic body frame. Radiother Oncol 2000;57:225–36. [11] Yorke E, et al. Dosimetric considerations in radiation therapy of coin lesions of the lung. Int J Radiat Oncol Biol Phys 1996;2:481–7.
Extracranial stereotactic RT
Comparison of two dose calculation methods applied to extracranial stereotactic radiotherapy treatment planning Anders Traberg Hansena,*, Jørgen B. Petersena, Morten Høyerb, Jens Juul Christensena a
Department of Medical Physics, and bDepartment of Oncology, Aarhus University Hospital, Aarhus, Denmark
Abstract Background and purpose: Lung tissue is a special challenge for a dose calculation algorithm, especially in the case of extracranial stereotactic radiotherapy (ESRT) due to small field sizes in combination with large variations in tissue density. The present study investigates the choice of dose calculation algorithm for 18 patients with a single lung tumor and 8 patients with a single liver tumor. The dose calculation is performed with both the pencil beam convolution algorithm and the collapsed cone convolution algorithm with the same number of monitor units in both cases. In addition, the dose calculation with the collapsed cone convolution algorithm is also performed with modified field sizes in order to match the Planning Target Volume (PTV) peripheral dose of the pencil beam based treatment. Results: For liver tumors, the mean Clinical Target Volume (CTV) dose calculated by the collapsed cone convolution algorithm and the pencil beam convolution algorithm is almost identical. For lung tumors, the mean CTV dose determined by the collapsed cone convolution algorithm differs up to 20%. Plans obtained by the two algorithms have field sizes which differ up to 8 mm for the same number of monitor units and minimum dose to the lung PTV. Conclusions: The choice of dose calculation algorithm can have a large influence on a treatment plan for ESRT of the lungs. q 2005 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 77 (2005) 96–98. Keywords: Extracraniel stereotactic radiotherapy; Treatment planning; Dose calculation algorithms; Collapsed cone
In recent years, several radiotherapy centers have introduced extracranial stereotactic radiotherapy (ESRT) as a new modality for treatment of tumors in the liver and in the lung [3,4,8,10]. SRT is characterized by the use of several small fields for the treatment of small volumes with a high dose [2,4]. This is a challenge for the treatment planning system (TPS), especially in the lung, where the scatter conditions are very different to those in water. In this study, the most commonly used calculation algorithm, the pencil beam convolution algorithm, is compared to the collapsed cone convolution algorithm, which is generally accepted to be more accurate [1,6,7,11].
Materials and methods The patients that this study is based on are immobilized by the Electa stereotactic body frame [4]. CT scans of the patients are carried out with a slice thickness of 5.0 mm and the subsequent dose planning and dose calculation is performed by the HELAX-TMS TPS version 6.1 B. The CTV is set equal to the delineated Gross Tumor Volume (GTV) and the PTV is obtained by expanding the
boundary of the CTV. The centrally located ICRU point [5] is usually coinciding with the treatment isocenter. The PTV is included in the 67% isodose, and the CTV is included in the 95% isodose (Lax originally prescribed a dose of 100% to the PTV periphery, corresponding to a dose of approximately 150% in the central part of the tumor, which is similar to prescription methods used in cranial stereotactic radiosurgery) [4]. The treatment plans consist of 4–8 photon fields with an energy of 6 MV in either a coplanar or a noncoplaner field arrangement. The field sizes used range from 4!4 to 10!10 cm and distance between the calculation points is 4.5 mm and the fields are shaped by a Siemens MLC with a leaf width of 1 cm at the isocenter. All patients are treated with plans calculated by the pencil beam convolution algorithm. For each patient, the treatment plan is recalculated using the collapsed cone convolution algorithm with the same number of monitor units as that of the pencil beam convolution algorithm. All beam related parameters are the same for the two calculation methods. During the initial planning of a patient, the MLC-leaves are automatically adapted to a contour obtained from an expansion of the projected PTV boundary,
0167-8140/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2005.04.018
A. Traberg Hansen et al. / Radiotherapy and Oncology 77 (2005) 96–98
Results and discussion Fig. 1 shows the ratios between the minimum PTV dose and the mean GTV dose calculated by the pencil beam or the collapsed cone convolution algorithms for lung tumors. It is observed that for some of the small tumors, the minimum doses to the PTV are up to 65% larger for the pencil beam convolution algorithm than for the collapse cone convolution algorithm. The mean dose to the GTV is best for reporting dose, since this quantity have a small dependence of the used algorithm and therefore suitable for this purpose. The same analysis for the 8 patients with liver tumors shows that the doses calculated by the two calculation algorithms are within 1% from each other. Fig. 2 shows the effect of the different margins used. It is observed that an additional margin around the PTV of up to 4 mm is needed for the same number of monitor units to obtain the same minimum dose to the PTV for the collapsed cone convolution algorithm as that of the pencil beam convolution algorithm. The collapsed cone convolution algorithm takes the range of the scattered secondary
PP-cal/CC-cal
1.8
1.6
PTV min
1.4
GTV mean
1.2
1
0.8 2
5
13
23 38 48 70 GTV volume (ccm)
91
190
Fig. 1. Ratio of minimum dose in the lung PTV calculated with the pencil beam convolution algorithm and the collapsed cone convolution algorithm for the 18 lung tumors (†), and the corresponding ratio for the mean dose in the GTV (;).
25 48 ccm 27 ccm
15 Relative deviation (%)
and subsequently adjusted to cover the PTV by the 67% isodose curve. To examine the effect of adding margins to the PTV, a systematic adaptation of MLC-leaves is used. The MLC-leaves of the initial treatment plan are therefore automatically readapted to the contour in such a way that the midpoints of the leaves are placed on the contour. This plan is evaluated by the pencil beam convolution algorithm. Additional plans are made in the same way except that the MLC-leaves are adapted to the contour of the projected PTV boundary with an extra margin of 0–4 mm. These plans are evaluated by the collapsed cone convolution algorithm and renormalised to the same number of monitor units as that of the plan used for treatment, and the minimum dose to the PTV is recorded. This procedure is performed for five selected patients with lung tumors.
97
141 ccm 2 ccm
5
5 ccm -5
-15
-25 0
1
2 3 Extra margin (mm)
4
Fig. 2. Difference in the minimum dose to the PTV between several collapsed cone calculations and a pencil beam calculation as a function of the extra margin from the projected PTV contour to the field edge for five lung tumors.
electrons into account [1]. This is especially important in the lung, where the range of secondary electrons is more than twice that of water. This difference in the handling of the scattered electrons has profound implications on the algorithms ability to describe the penumbra region. It is therefore generally accepted that the collapsed cone convolution algorithm is superior to the pencil beam convolution algorithm for the calculation of radiation dose in lung tissue [1,3,6,7,11]. The pencil beam convolution algorithm tends to overestimate the dose to a small lung tumor in comparison to the collapsed cone convolution algorithm, as shown in Fig. 1, because the loss of scattered particles is significant for a small tumor. For a dose calculation in the liver tissue, the two algorithms predict almost the same doses, since the liver is more equivalent to water. In ESRT, the dose delivered to the tumor volume is not necessarily homogeneous. It has therefore been suggested to focus on the minimum dose delivered to the PTV [10]. However, in the case of a lung tumor, the minimum dose is typically located at the surface of the PTV, where there is a steep dose gradient. This makes the minimum dose to the PTV very sensitive to the choice of dose calculation algorithm. It is also possible to quantify the differences between the calculation algorithms by keeping the field weights, field arrangement, and monitor units fixed. A plan with a specified minimum dose to the PTV based on the pencil beam convolution algorithm can be reproduced by the collapsed cone convolution algorithm by increasing the margins, and by using the same number of monitor units. Instead of claiming that the pencil beam convolution algorithm overestimates doses to lung tumors, it can as well be claimed the pencil beam convolution algorithm underestimates the margins needed to give a prescribed minimum dose to the PTV. This study reveals that the underestimation of margins in some cases exceeds 4 mm. Omitted by this study are other sources of uncertainty in the process of ESRT for instance the important issue of tumor movement [9,10].
98
Comparison of two dose calculation methods
In conclusion, we find that the pencil beam convolution algorithm tends to overestimate the dose to small lung tumors relative to the collapsed cone convolution algorithm, which is corresponding to an underestimation the margins needed to secure a sufficient coverage of the PTV. Therefore, by using the pencil beam convolution algorithm, one could under dose small lung tumors significantly. We therefore recommend that a collapsed cone calculation is used for treatment of lung tumors. In addition, dose prescription to the periphery of a lung tumor should be avoided.
*
Corresponding author. Anders Traberg Hansen, Tel.: 89492594; fax: 89492590. E-mail address: [email protected] Received 25 May 2004; received in revised form 5 April 2005; accepted 6 April 2005
References [1] Aspradakis MM, Morrison RH, Richmond ND, Steele A. Experimental verification of convolution/superposition photon dose calculations for radiotherapy treatment planning. Phys Med Biol 2003;48:2873–93. [2] Brahme A. Dosimetric precision requirements in radiation therapy. Acta Radiol Oncol 1984;23:379–91.
[3] Butson MJ, et al. Verification of lung dose in an anthropomorphic phantom calculated by the collapsed cone convolution method. Phys Med Biol 2000;45:N143–N9. [4] Ingmar Lax, Henric Blomgren, Ingemar Na ¨slund, Rut Svanstro ¨m. Stereotactic radiotherapy of malignancies in the Abdomen. Acta Oncologica 1994;33:677–83. [5] ICRU (International Commission on Radiological Units) Report 50. Prescribing, recording, and reporting photon beam therapy. Washington, DC: ICRU; 1993. [6] Lydon JM. Photon dose calculations in homogeneous media for a treatment planning system using a collapsed cone superposition convolution algorithm. Phys Med Biol 1998;43:1813–22. [7] Weber L, Nilsson P. Verification of Dose Calculations with a Clinical Treatment Planning system based on a Point Kernel dose engine. J Appl Clin Med Phys 2002;3(2):73–87. [8] Wulf J, et al. Stereotactic radiotherapy for primary lung cancer and pulmonary metastases: a noninvasive treatment approach in medically inoperable patients. Int J Radiat Oncol Biol Phys 2004;60:186–96. [9] Wulf J, et al. Impact of target reproducibility on tumor dose in stereotactic radiotherapy of targets in the lung and liver. Radiother Oncol 2003;66:141–50. [10] Wulf J, et al. Stereotactic radiotherapy of extracranial targets: CT-simulation and accuracy of treatment in the stereotactic body frame. Radiother Oncol 2000;57:225–36. [11] Yorke E, et al. Dosimetric considerations in radiation therapy of coin lesions of the lung. Int J Radiat Oncol Biol Phys 1996;2:481–7.