Optimization of extracranial stereotactic radiation therapy of small lung lesions using accurate dose calculation algorithms
 Barbara Dobler^{1, 2}Email author,
 Cornelia Walter^{1},
 Antje Knopf^{1},
 Daniella Fabri^{1},
 Rainer Loeschel^{3},
 Martin Polednik^{1},
 Frank Schneider^{1},
 Frederik Wenz^{1} and
 Frank Lohr^{1}
DOI: 10.1186/1748717X145
© Dobler et al; licensee BioMed Central Ltd. 2006
Received: 27 September 2006
Accepted: 29 November 2006
Published: 29 November 2006
Abstract
Background
The aim of this study was to compare and to validate different dose calculation algorithms for the use in radiation therapy of small lung lesions and to optimize the treatment planning using accurate dose calculation algorithms.
Methods
A 9field conformal treatment plan was generated on an inhomogeneous phantom with lung mimics and a soft tissue equivalent insert, mimicking a lung tumor. The dose distribution was calculated with the Pencil Beam and Collapsed Cone algorithms implemented in Masterplan (Nucletron) and the Monte Carlo system XVMC and validated using Gafchromic EBT films. Differences in dose distribution were evaluated. The plans were then optimized by adding segments to the outer shell of the target in order to increase the dose near the interface to the lung.
Results
The Pencil Beam algorithm overestimated the dose by up to 15% compared to the measurements. Collapsed Cone and Monte Carlo predicted the dose more accurately with a maximum difference of 8% and 3% respectively compared to the film. Plan optimization by adding small segments to the peripheral parts of the target, creating a 2step fluence modulation, allowed to increase target coverage and homogeneity as compared to the uncorrected 9 field plan.
Conclusion
The use of forward 2step fluence modulation in radiotherapy of small lung lesions allows the improvement of tumor coverage and dose homogeneity as compared to nonmodulated treatment plans and may thus help to increase the local tumor control probability. While the Collapsed Cone algorithm is closer to measurements than the Pencil Beam algorithm, both algorithms are limited at tissue/lung interfaces, leaving MonteCarlo the most accurate algorithm for dose prediction.
Background
In external beam radiation therapy of lung tumors, even in extracranial stereotactic high dose ablative treatments, infield relapses are observed even at high doses [1–3].
It is well known that the Pencil Beam dose calculation algorithm (PB), which is most commonly implemented in commercial treatment planning systems, has limited accuracy especially at the interface of tissues with large differences in electron density, as it is the case at the interface lung/tumor [4]. Collapsed Cone convolution methods (CC) and Monte Carlo simulations (MC) are able to predict the dose distribution with a higher accuracy [5–10]. They are, however, more time consuming and therefore implemented in a few commercial treatment planning systems only and not widely used in clinical routine yet.
Several studies have been published, which report the differences in dose distributions in inhomogeneous media predicted by the different dose calculation algorithms. Martens et al [8] reported an overestimation of the dose in the upperairway mucosa by the Pencil Beam algorithm. Haedinger et al [10] and Koelbl et al [9] found that the Pencil Beam algorithm overestimates the dose to targets in the lung as compared to the Collapsed Cone algorithm, which may lead to an insufficient dose to the target volume. These reports, however, do not include any 2D dose measurements in anthropomorphic phantoms or investigations about how the treatment plans can be improved in order to achieve the prescribed dose at the periphery of the target without increasing lung toxicity.
The aim of this study was to optimize the treatment of small lung tumors with regard to target coverage and dose homogeneity using accurate dose calculation algorithms and to validate the results in an anthropomorphic phantom. The first step was to quantify the error introduced when creating plans using the Pencil Beam dose calculation algorithm by recalculating plans with Collapsed Cone, Monte Carlo and comparing them to absolute dose measurements in an inhomogeneous thoracic phantom. Once the error was quantified, the plans were optimized by adding small segments to the peripheral parts of the target, creating a 2step fluence modulation. Similar to what is done in inverse planning of IMRT, this fluence modulation is used to create a steeper dose gradient at the interface of the target to the lung, in order to increase the dose to the periphery of the target while keeping the dose to the lung low. To achieve a homogeneous dose distribution for dose escalation in a larger volume of the target, the weights of the segments were altered to determine the optimal weights for the desired dose homogeneity using Collapsed Cone and Monte Carlo calculations. The calculations were validated by 2D film dosimetry.
Methods
Phantom Design
Dose Calculation Algorithms
Dose calculations were performed using the Pencil Beam algorithm and the Collapsed Cone algorithm implemented in Masterplan (Nucletron BV, Venendal, the Netherlands). The algorithms are described in detail by Ahnesjö et al [11, 12]. For the Monte Carlo simulations XVMC including the Elekta Synergy treatment head model was kindly supplied by Matthias Fippel [13, 14]. A detailed overview over the different dose calculation methods has been given by Ahnesjö and Aspradakis [15].
Treatment Plans
Dosimetric Validation
Dosimetric validation was performed by absolute film dosimetry using the Gafchromic^{®} EBT radiochromic film (ISP, New Jersey, USA) and the EPSON 1680 scanner [16–20]. The properties of the film in combination with the EPSON 1680 scanner were thoroughly investigated prior to the use in this study. The overall error in absolute dose is highly dependent on the film handling and scanner settings and can be as much as 40% if inappropriate settings are used. Sources of error are e.g. the use of an automatic color correction, the increase of scanner temperature and exceeding exposure of the film to light. The use of proper scanner settings and proper film handling, however, allow to reduce the maximum error to 3% [21] in the central region (10 cm × 20 cm) of the scanner. In the periphery of the scanning field, i.e. more than 10 cm from the center, the error can be as much as 8% without applying further corrections due to the horizontal inhomogeneity of the scanner [22].
The setup error of the phantom at the linac was estimated as ± 1 mm. The influence of the setup error on the measured dose distribution in craniocaudal direction was eliminated by using a phantom geometry which does not change in the craniocaudal direction (cylindrical targets). Fore spherical targets, even small setup errors in the craniocaudal direction cause a change in target diameter in the transversal plane. This would have added an additional uncertainty in the evaluation which could be eliminated by using cylindrical targets.
Moving Targets
To test target coverage for moving targets, moving targets were simulated by shifting the isocenter relative to the phantom by 5 mm in the lateral direction and recalculating the plan. Even though the main direction of target movement is in the craniocaudal direction, the lateral direction was used because there is no fluence modulation in craniocaudal direction and lateral movement of our cylindrical insert also mimics craniocaudal movement of a clinical lesion that is typically spherical.
Evaluation
The evaluation of the accuracy of dose calculation was performed by comparison of dose profiles and difference matrices of the calculated and measured doses. Evaluation of plan optimization was performed by means of Dose Volumes Histograms (DVHs) calculated by CC and PB and comparison of dose profiles of the calculated and measured dose matrices. DVHs of MC calculated plans are not available since our version of XVMC does not support dose volume histograms.
Results
Comparison of dose calculation algorithms for the 9 field plan
Plan optimization
Moving targets
Discussion
The goal of the study was to optimize high dose precision radiation therapy of small lung lesions using accurate dose calculation algorithms. In a first step, a nonmodulated 9 field conformal treatment plan was created on an inhomogeneous phantom, calculated with Pencil Beam, Collapsed Cone, and Monte Carlo, and validated by 2D film measurements. Then the plan was optimized by adding small segments to the peripheral part of the target in the sense of a twostep intensity modulation in order to increase target coverage and dose homogeneity without compromising the dose to the lung.
Accuracy of dose calculation
The measurements showed a very good agreement with the Monte Carlo dose calculation, with a maximum difference of up to 3%, which is in within the accuracy of film dosimetry of ± 3%. The Pencil Beam algorithm overestimated the dose compared to the film by up to 15% at the interface of the target to the lung, whereas the Collapsed Cone underestimated the dose in this region by up to 8%.
These results confirm the accuracy of the MC calculations and the overestimation of the dose by the Pencil Beam algorithm as it was published in several studies [5–10]. This can be explained by the fact that PB calculations do not model the rebuildup correctly because electron transport is not taken into account. As well, only a one dimensional longitudinal inhomogeneity correction is performed whereas lateral tissue heterogeneities are neglected.
The underestimation of the dose by the Collapsed Cone algorithm, however, has not been so clearly stated until now. Aspradakis et al [23] reported that the monitor units calculated with CC were within ± 2% of the measurements. For tissue heterogeneities the maximum deviation was 3.8% at the central axis for a large lung block of 25 cm × 25 cm × 10 cm in a large water equivalent block of 50 cm × 50 cm × 20 cm. In the penumbra region, however, a maximum dose deviation of 28% could be found. For missing backscatter the error was up to 5%. Weber and Nilsson [24] reported good agreement of the Collapsed Cone algorithm with measurements in most cases. However, calculations were outside the limits of 5% in the buildup region.
Haedinger et al [10] found an average difference in absolute dose of 5.4% (SD 5.8%) for the Collapsed Cone algorithm compared to the Pencil Beam algorithm, increasing with decreasing sizes of the CTV (2 cm^{3} to 256 cm^{3}). The shape of the MC profiles resembled the Collapsed Cone calculations more closely, however, only relative and no absolute profiles were compared, and no measurements were performed. Koelbl et al [9] performed measurements with a 6MV photon beam in a phantom consisting of slices of styrofoam (lung parenchyma) and RW3 (chest wall). The beam direction was parallel to the tissue interfaces. In this setup, the dose measured with an ion chamber at 15 cm depth in the tumor tissue part of the phantom showed a difference of only 1.2% compared to the CC calculations.
The relatively high differences between the dose calculated by Collapsed Cone and the measured dose found in our study can be explained by the fact that the calculations were performed for 9 fields with perpendicular incidence to the lung/tissue interface of a small lung lesion of 2 cm diameter and 4 cm height. In this case incorrect modeling of the secondary buildup as described by Weber and Nilsson [24] as well as missing backscatter as described by Aspradakis et al [23] have a large influence on the calculated dose.
Plan optimization
It is possible to improve target coverage by simply renormalizing the 9 field Collapsed Cone plan to the isocenter, thus increasing the number of MU and dose to the ipsilateral lung. Using additional peripheral segments, however, also the dose homogeneity can be improved and the dose to the ipsilatieral lung can be lowered. Lower weighting of the peripheral segments was not possible for a prescription dose of 2Gy because the linac handles only integer MU. For higher prescription doses as used in clinical practice, even lower weighting of the segments might be worth considering.
Moving Targets
In principle the technique is applicable to moving targets also. In this case the PTV margins have to be increased in the craniocaudal direction, which is the main direction of the movement of targets in the lung. The field size of the additional segments as well as of the open fields have to be increased in craniocaudal direction accordingly.
If additional segments are used, the dose gradient gets steeper. If the dose is now prescribed to the same isodose encompassing the PTV, the part of the lung which is outside the PTV will get a lower dose, but the part of the lung inside the PTV will get a higher dose. It is therefore recommended to use breath hold or gating techniques to be able to keep PTV margins as small as possible.
Since the lung density changes with exhalation and inhalation, the dose distribution can be modeled correctly only if the treatment is performed in the same status of exhalation or inhalation as the planning CT [25]. The lowest dose to the lung will be achieved for deep inhalation, since the percentage of irradiated lung volume decreases as the overall lung volume increases.
Conclusion
The use of forward 2step intensity modulation in radiotherapy of small lung lesions allows the improvement of tumor coverage and dose homogeneity with a somewhat lower dose to the lung compared to nonmodulated treatment plans and may thus help to increase the local tumor control probability. While the Collapsed Cone algorithm agrees better with measurements than the Pencil Beam algorithm, both algorithms have limited accuracy at tissue/lung interfaces. Since this is especially problematic for small targets, MonteCarlo methods should be used in this situation. Breath hold or gating techniques are recommended in order to keep the PTV margins small.
List of Abbreviations
 2D:

2 Dimensional
 PB:

Pencil Beam Algorithm
 CC:

Collapsed Cone Algorithm
 CTV:

Clinical Target Volume
 DVH:

Dose Volume Histogram
 MC:

Monte Carlo
 PTV:

Planning Target Volume
Declarations
Acknowledgements
We want to thank Matthias Fippel for supplying XVMC including the Elekta Synergy treatment head model.
Authors’ Affiliations
References
 Hof H, Herfarth KK, Munter M, Hoess A, Motsch J, Wannenmacher M, Debus JJ: Stereotactic singledose radiotherapy of stage I nonsmallcell lung cancer (NSCLC). Int J Radiat Oncol Biol Phys 2003, 56: 335341. 10.1016/S03603016(02)045042View ArticlePubMedGoogle Scholar
 Timmerman R, Papiez L, McGarry R, Likes L, DesRosiers C, Frost S, Williams M: Extracranial stereotactic radioablation: results of a phase I study in medically inoperable stage I nonsmall cell lung cancer. Chest 2003, 124: 19461955. 10.1378/chest.124.5.1946View ArticlePubMedGoogle Scholar
 Wulf J, Hadinger U, Oppitz U, Thiele W, Flentje M: Impact of target reproducibility on tumor dose in stereotactic radiotherapy of targets in the lung and liver. Radiother Oncol 2003, 66: 141150. 10.1016/S01678140(02)003729View ArticlePubMedGoogle Scholar
 Knoos T, Ahnesjo A, Nilsson P, Weber L: Limitations of a pencil beam approach to photon dose calculations in lung tissue. Phys Med Biol 1995, 40: 14111420. 10.1088/00319155/40/9/002View ArticlePubMedGoogle Scholar
 Osei EK, Darko J, Mosseri A, Jezioranski J: EGSNRC Monte Carlo study of the effect of photon energy and field margin in phantoms simulating small lung lesions. Med Phys 2003, 30: 27062714. 10.1118/1.1607551View ArticlePubMedGoogle Scholar
 McDermott PN, He T, DeYoung A: Dose calculation accuracy of lung planning with a commercial IMRT treatment planning system. J Appl Clin Med Phys 2003, 4: 341351. 10.1120/1.1623172View ArticlePubMedGoogle Scholar
 Butson MJ, Elferink R, Cheung T, Yu PK, Stokes M, Quach KY, Metcalfe P: Verification of lung dose in an anthropomorphic phantom calculated by the collapsed cone convolution method. Phys Med Biol 2000, 45: N1439. 10.1088/00319155/45/11/402View ArticlePubMedGoogle Scholar
 Martens C, Reynaert N, De Wagter C, Nilsson P, Coghe M, Palmans H, Thierens H, De Neve W: Underdosage of the upperairway mucosa for small fields as used in intensitymodulated radiation therapy: a comparison between radiochromic film measurements, Monte Carlo simulations, and collapsed cone convolution calculations. Med Phys 2002, 29: 15281535. 10.1118/1.1487421View ArticlePubMedGoogle Scholar
 Koelbl O, Krieger T, Haedinger U, Sauer O, Flentje M: Influence of Calculation Algorithm on Dose Distribution in Irradiation of NonSmall Cell Lung Cancer (NSCLC)Collapsed Cone Versus Pencil Beam. Strahlenther Onkol 2004, 180: 783788. 10.1007/s0006600412684View ArticlePubMedGoogle Scholar
 Haedinger U, Krieger T, Flentje M, Wulf J: Influence of calculation model on dose distribution in stereotactic radiotherapy for pulmonary targets. Int J Radiat Oncol Biol Phys 2005, 61: 239249. 10.1016/j.ijrobp.2004.03.028View ArticlePubMedGoogle Scholar
 Ahnesjo A: Collapsed cone convolution of radiant energy for photon dose calculation in heterogeneous media. Med Phys 1989, 16: 577592. 10.1118/1.596360View ArticlePubMedGoogle Scholar
 Ahnesjo A, Saxner M, Trepp A: A pencil beam model for photon dose calculation. Med Phys 1992, 19: 263273. 10.1118/1.596856View ArticlePubMedGoogle Scholar
 Fippel M: Fast Monte Carlo dose calculation for photon beams based on the VMC electron algorithm. Med Phys 1999, 26: 14661475. 10.1118/1.598676View ArticlePubMedGoogle Scholar
 Fippel M, Haryanto F, Dohm O, Nusslin F, Kriesen S: A virtual photon energy fluence model for Monte Carlo dose calculation. Med Phys 2003, 30: 301311. 10.1118/1.1543152View ArticlePubMedGoogle Scholar
 Ahnesjo A, Aspradakis MM: Dose calculations for external photon beams in radiotherapy. Phys Med Biol 1999, 44: R99155. 10.1088/00319155/44/11/201View ArticlePubMedGoogle Scholar
 Devic S, Seuntjens J, Sham E, Podgorsak EB, Schmidtlein CR, Kirov AS, Soares CG: Precise radiochromic film dosimetry using a flatbed document scanner. Med Phys 2005, 32: 22452253. 10.1118/1.1929253View ArticlePubMedGoogle Scholar
 Butson MJ, Cheung T, Yu PK: Absorption spectra variations of EBT radiochromic film from radiation exposure. Phys Med Biol 2005, 50: N13540. 10.1088/00319155/50/13/N02View ArticlePubMedGoogle Scholar
 Butson MJ, Cheung T, Yu PK: Weak energy dependence of EBT gafchromic film dose response in the 50kVp10MVp Xray range. Appl Radiat Isot 2006, 64: 6062. 10.1016/j.apradiso.2005.07.002View ArticlePubMedGoogle Scholar
 Cheung T, Butson MJ, Yu PK: Postirradiation colouration of Gafchromic EBT radiochromic film. Phys Med Biol 2005, 50: N2815. 10.1088/00319155/50/20/N04View ArticlePubMedGoogle Scholar
 Rink A, Vitkin IA, Jaffray DA: Characterization and realtime optical measurements of the ionizing radiation dose response for a new radiochromic medium. Med Phys 2005, 32: 25102516. 10.1118/1.1951447View ArticlePubMedGoogle Scholar
 Schneider F, Polednik M, Wolff D, Steil V, Delana A, Lohr F, Wenz F, Menegotti L: Evaluation von GafchromicEBTFilmen für die IMRTPlanverifikation. Strahlenther Onkol 2006,182( (Sondernr. 1)):99.Google Scholar
 Menegotti L, Delana A, Martignano A, Vanoni V, Mazzucchi S: 2D dosimetry with Gafchromic EBT: IMRT planning verification. Radiotherapy & Oncology 2006,81(supplement 1):363.Google Scholar
 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: 28732893. 10.1088/00319155/48/17/309View ArticlePubMedGoogle Scholar
 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: 7387. 10.1120/1.1451983View ArticlePubMedGoogle Scholar
 Aarup L, Korreman S, Nyström H, JuhlerNoettrup T, Specht L, Jutemark B, Nahum AE: A phantom study comparing radiation dose differences due to different dose calculation algorithms as a function of lung density. Radiotherapy & Oncology 2006,81(supplement 1):99.Google Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.