Investigation of AAA dose calculation algorithm accuracy in surface and buildup region for 6MV photon beam using markus parallel-plate ion chamber

Abstract

In radiotherapy, dose distributions are obtained by using dose calculation algorithms that are implanted in treatment planning systems (TPS). This study aims to compare the surface doses of separate field sizes calculated by different version of The Analytical Anisotropic Algorithm (AAA) and measured by the parallel-plate ion chamber that is admitted as the most reliable dosimetry system for the surface region dose measurements. In order to measure the near surface dose, water equivalent solid phantom was used and measurements were made for 6MV photon beam at 100 cm source-detector distance for 5×5, 10×10, and 20×20 cm² field sizes. AAA 8.9 and AAA 15.1 versions of the Varian Eclipse TPS were used for surface dose calculations by generating beams with separate field sizes. The doses were read by considering the effective buildup thickness of Markus parallel-plate ion chamber. The surface doses using 6 MV photon beams for 10×10 cm² field size at 0.07 mm were found to be 11.04%, 26.25%, and 19.69% for AAA v8.9, AAA v15.1 and Markus chamber, respectively. It was seen that for both of the AAA versions and Markus parallel-plate ion chamber, increasing field sizes also increase surface dose. For all field sizes, surface dose was lowest by using AAA v8.9 at 0.07 mm. The different versions of the same TPS algorithms may calculate the surface doses distinctively. After upgrading of TPS algorithms, surface doses should be calculated and compared by measurements with different dosimetry systems to better understand their calculation behaviors in the near surface region.

Keywords

AAA dose algorithm surface dose buildup region Markus parallel plate ion chamber

1. Introduction

In external radiotherapy, the computerizing treatment planning systems (TPS) are used to generate beam shapes and dose distributions in order to provide maximum tumor control while minimizing critical organ doses. Developments in computer systems have enabled the development of computerized tomography devices simultaneously. By the introduction of CT-based planning systems into the radiotherapy world, it is available to obtain patient anatomy, target volume and dose distributions as three-dimensional (3D) views. In radiation therapy the accuracy of the dose delivered to the patient is a very crucial issue. This issue is mostly affected by the accuracy of treatment planning dose calculation algorithms.

Dose calculation algorithms required by treatment planning computers have been developing since the middle of the 1950 s. Some recommendations about the verification and commissioning of dose calculation algorithms being used in TPS are given by different reports such as AAPM TG-53 IAEA TRS 430 and NCS Report 15 [1].

There are some requirements about treatment planning dose calculation. They must be fast enough in order to complete the treatment planning process within the clinically acceptable times and simultaneously the dose calculation results should have enough accuracy. Due to the necessity of the detailed energy transport information in the patient, to obtain an accurate dose calculation in heterogeneous tissues is also an important issue. Simple dose calculation methods can be used for calculation of clinical cases which consist of homogenous tissues. All new practices related to dose algorithms concern about to improve the dose calculation accuracy in the heterogeneous tissues such as around the air cavities [2]. Treatment planning dose calculation algorithms can be separated into three categories; Correction-based algorithms, Model-based algorithms and direct Monte Carlo [1, 3].

Correction-based algorithms are the easiest dose calculation methods but also, they give the least accurate results. The calculations can be made by manually in the correction based models. Tissue Air Ratio (TAR) Method, Batho Power Law Method, Equivalent Tissue Air Ratio (ETAR) method and Isodose Shift Method are some of the methods that they can be used for manual calculations as well as made part of a correction-based algorithm for the calculation of absorbed dose at a point in a patient [3].

Model-Based Algorithms are more complex than Correction-Based Algorithms and must be performed by high performance computers. The use of algorithms and TPSs for 3D and intensity modulated radiation therapy is based on early 1990s. The problems in the traditional dose calculations occurred from tissue inhomogeneities and absorbed-dose differences due to varying field shape geometries were overcame by model-based algorithms. These algorithms were based on convolution or superposition methods.

The most accurate method of calculating patient dose distribution is Monte Carlo. This technique is a computer program which simulates huge number of photons and particles transportation through the matter and determines the probability of individual interactions of them using the fundamental physics laws. The number of simulated particles effect the accuracy of their predicting dose distribution. The greater the number, the greater the accuracy of the estimated dose distribution. However, it takes quite a long time to simulate particles in excessive amounts. For this reason, their usage in clinics is limited. Today it is mostly used for experimental and research purposes.

Today, many commercial clinical treatment planning systems use model-based dose calculation algorithms instead of correction-based. Although there are great advances in dose calculation algorithms, it is still a critical issue to predict the accurate skin doses by the treatment planning systems. It is extremely problematic to make an accurate calculation in buildup region where the charge particle equilibrium is not established. The limitations of model-based dose calculation algorithms at the buildup region are still exploring by many research groups. Wang et al. reported that Eclipse treatment planning system with AAA model-based calculation algorithm generally underestimated skin doses by up to 14% [4].

Oinam et al. arranged a seven field IMRT plan on a phantom by using AAA and Pencil Beam algorithms. Their results showed that both algorithms have limitations on predicting build-up dose accurately at depth less than 2 mm. The surface dose calculated by the AAA algorithm was reported to be 7.56 % lower than measurement results [5].

In order to fix the errors of the previous version and obtain more accurate dose calculation, TPS manufacturers update their treatment planning system versions at specific time intervals. The vendor gives training to the customers about how the new version of the system is working but commonly the detailed information about changes in the new software version is not given to the customer.

Even though, in the literature, while comparative studies of different versions of the Eclipse TPS have been done by various groups, but there is not any study about their behaviors in the surface region. For example, Krishna et al. quantified the dose difference between two versions of AAA algorithms from 8.8 to 13.6 version. For this purpose, the therapy plans of five sites which belongs to head and neck, breast, lung, cervix and stomach recalculated and reevaluated by new algorithm version. The therapy plans were compared in terms of different indices such as homogeneity or conformity indexes. It was reported that, no significant difference was found between old and new versions according to the planning target volume coverage and sparing of critical organs [6].

In the previous studies, it is reported that there are differences in surface doses, although there is no difference in tumor coverage between different versions. These surface dose differences should be measured according to the version.

The most accurate buildup dose can be measured by extrapolation ion chambers but not every institute has it. Although they need some correction factors, parallel-plate ion chambers can be a good alternative to extrapolation ion chamber for buildup dose measurements.

Purpose of this study was to investigate the change of the buildup dose depending on the version differences by taking the Markus parallel-plate ion chamber measurements as a reference.

2. Materials and methods

2.1. Parallel plate ionization chamber measurement

In this study, a Markus parallel-plate ionization chamber was used for surface and buildup region dose measurements. The surface dose was defined at the depth of 0.07 mm as it is recommended by The International Commission on Radiation Units and Measurements (ICRU) and the International Commission on Radiological Protection (ICRP)⁷. Markus chamber has an effective measurement point at 0.023 mm.

The fixed plane separation of Markus chamber is 2 mm and the distance from sidewall to collector is 0.35 mm. Unidose dosimeter (PTW Freiburg, Germany) was performed to acquire the relative dose at any point in the buildup region. The readings of the chamber were corrected due to the polarity effect by the formula: Q_avg = (Q₊ + Q_–)/2. Q_avg is the average charge where Q+ and Q_– are the positive and negative polarities, respectively.

A set of water equivalent RW3 slab phantoms (SP34, PTW Freiburg, Freiburg, Germany) which has less uncertainty compare to liquid form of water was used for open field irradiation. The surface area of the phantom set was 40×40 cm² and it has a thickness of 15 cm (Each slab phantom is 1 cm thick). The phantom set has a physical density of 1.045 gcm^–3. The physical density of the phantom was considered to acquire the water equivalent depth (WED). The effective measurement point of the chamber was defined by using WED values in the buildup region.

Surface and buildup region dose measurements were carried out with a Markus chamber for 5×5, 10×10 and 20×20 cm² field sizes at 0, 1, 2, 3, 4, 5, 10 and 15 mm phantom depths. Figure 1 shows the measurement setup for Markus parallel plane ion chamber.

Irradiations were made using a Varian Trilogy linear accelerator (Varian, Palo Alto, CA) with 6 MV photon energy in a fixed source to skin distance (SSD) of 100 cm by delivering 100 MU. To obtain the percentage depth doses (PDDs), the measured doses were then normalized to the dose at 15 mm phantom depth where was accepted the dose maximum depth. The overdoses obtained from Markus chamber were corrected by applying Gerbi’s method to PDDs [7]. The measured dose values at the phantom depths were used for interpolation calculations to obtain the doses at 0.07 mm WED. The same interpolation calculation was performed to acquire the doses at WEDs of the first five millimeters in buildup region.

Fig.1

Measurement setup of markus parallel plane ion chamber with different plate thicknesses.

2.2. TPS calculations

Manufacturers upgrade their versions in certain periods in order to eliminate previous version errors, achieve more accurate results in dose calculations and add user-friendly options to the planning systems. In our clinic, 8.9 version of AAA algorithm which was a very old version of Eclipse TPS was upgraded to 15.1 version by recommendation of manufacturer. According to the manufacturer, compared to the version 8.9, 15.1 version provides a better point dose calculation accuracy and irregular treatment planning. In addition, the new version uses voxel-based calculations while the earlier version uses point dose calculations. Depending on the versions, such characteristics may cause differences in dose calculations.

The algorithm performed for calculations was AAA which is a model-based algorithm. AAA algorithm, which has been implemented in Eclipse TPS is one of a 3D pencil beam convolution-superposition algorithm. AAA model configuration is based on basic physical parameters which are determined by Monte-Carlo. This algorithm model calculates tissue heterogeneities anisotropically by using 13 lateral photon scatter kernels. Superposition of the doses determined from photon and electron convolutions gives the final dose distribution. The AAA dose calculation model has been developed in order to provide a fast Monte-Carlo-based 3D convolution/superposition algorithm and to obtain accurate heterogeneity-corrected photon dose calculation. It is originally developed by Drs. Waldemar Ulmer [8] and Wolfgang Kaissl [9].

The configuration algorithm and actual dose calculation algorithm are two main components of the AAA dose calculation model. The configuration algorithm determines the basic physical parameters. The fluence and energy spectra of the photons and electrons and also their scattering properties in the medium are characterized by the basic physical parameters. All the parameters are pre-computed by Monte-Carlo simulations in order to ensure the highly accurate results more quickly. After some procedures of the beam configuration phase are completed, actual dose calculation retrieved all parameters which are stored before [10].

There are no big differences in accuracy of calculation for the algorithms in homogenous media such as water but in the surface dose calculations, the point of calculation is located at the border of phantom surface and air where the heterogeneity effects the calculation. The AAA takes account of heterogeneity into longitudinal-lateral directions and it uses Gaussian functions to define the mean heterogeneous effect [11].

To assess the difference between the versions, first of all, the computed tomography (CT) images of the same set of phantom that used for Markus chamber measurements was acquired by using Philips Brilliance Big Bore CT (Philips Healthcare, Cleveland, OH) with a 3 mm of slice thickness. These images were then sent to the TPSs for surface and buildup region dose calculations. The modelled Varian Trilogy linear accelerator and 6 MV photon energy options were chosen in TPSs and the calculation grid was set to 2.5 mm. The irradiation fields with the same sizes and MU values as the Markus chamber measurements were generated in the TPSs for surface and buildup region dose calculations. After the calculations, the doses at the same points in chamber measurements were obtained from TPSs for comparison.

3. Results

The percentage depth doses at the surface and buildup regions for different field sizes measured using the Markus parallel-plate ionization chamber are given in Table 1. In order to obtain doses at 0.07, 1, 2, 3, 4 and 5 mm depth for Markus parallel-plate ion chamber, extrapolation was made by using four-order polynomial functions. The calculated doses were read in TPS for the same depths.

Table 1
Percentage depth doses (PDDs) obtained with a Markus chamber in a water equivalent RW3 phantom using 6 MV photon beams for different field sizes at SSD = 100 cm

Phantom WED PDD (%)

Depth (mm) (mm) 5×5 cm² 10×10 cm² 20×20 cm²

0 0,023 10,81 16,61 28,06

0,067 0,07 14,00 19,69 30,87

1 1,068 38,74 43,44 52,48

2 2,113 57,12 60,98 68,19

5 5,248 83,65 85,95 90,00

10 10,473 97,77 98,37 99,24

15 15,698 100,00 100,00 100,00

Phantom	WED	PDD (%)
0	0,023	10,81	16,61	28,06
0,067	0,07	14,00	19,69	30,87
1	1,068	38,74	43,44	52,48
2	2,113	57,12	60,98	68,19
5	5,248	83,65	85,95	90,00
10	10,473	97,77	98,37	99,24
15	15,698	100,00	100,00	100,00

The percentage depth doses in the near surface for 6 MV photon beams were calculated by Eclipse TPS v8.9 and v15.1 for the square field sizes of 5×5, 10×10 and 20×20 cm². The extrapolated skin dose values at 0.07 mm for Markus ion chamber are 19.69% for 10×10 cm². The calculated surface dose values for Eclipse TPS v8.9 and v15.1 are 11.04% and 26.25% for 10×10 cm². The percentage depth doses belong to different field sizes at 0.07 mm is shown in Fig. 2. As can be seen from Fig. 2, for all field sizes AAA v8.9 shows lowest results. For 5×5 and 10×10 cm², surface depth doses calculated by AAA v15.1 are higher than AAA v8.9 and Markus ion chamber.

Fig.2

Percentage depth doses varying with field sizes at 0.07 mm which is the recommended by ICRU as surface dose depth.

The dosimeters have their own specific effective measurement depth. Therefore, to obtain the doses by the dosimeter and TPS at the same depths, the WED of dosimeter must be considered. To achieve this manner, the PDD values in Table 1 was used to calculate the doses at the same effective depths in the buildup region. The PDD results for 5×5, 10×10, and 20×20 cm² field sizes are shown in Fig. 3.

Fig.3

Percentage depth doses (PDDs) for 6 MV photon beams with different sizes at 100 cm fixed SSD. The doses at 0, 1, 2, 3, 4, and 5 mm depths were obtained by markus parallel-plate ion chamber and Eclipse TPS with different versions of AAA algorithm, and the following field sizes were investigated: 5×5 cm² (a), 10×10 cm² (b), and 20×20 cm² (c).

4. Discussion

The dose measurement of surface and in the near surface region requires great attention in many radiotherapy applications [12, 13]. Firstly, the dosimetry preference is a challenging issue and their characteristics such as effective measurement depths should be known in great detail. Secondly, TPS cannot calculate the surface doses correctly. The most basic reason for this situation is concerned about how accurate the surface measurements to be used for beam modeling are, and which dosimetry system is used. Due to the skin toxicity in radiotherapy, accuracy of the surface dose calculation, especially within the initial 2 mm depth, has a great importance. In this study, the surface dose was measured with a parallel-plate ion chamber and calculated by v8.9 and v15.1 of AAA algorithm for different field sizes and different depths. Thus, the surface dose calculation accuracy of the different versions of the AAA algorithm was checked by comparing the calculated and measured surface dose values. Extrapolation chambers are the recommended dosimeters in order to obtain the most accurate results for surface region [14].

In our study, Markus parallel-plate ion chamber was used which is admitted as an alternative to the extrapolation chambers. After having used Gerbi’s overresponse correction factors, parallel-plate ion chambers can be used peace of mind. In this study, Markus parallel-plate ion chamber was utilized as a reference to evaluate the dose responses of different versions of AAA algorithm in the near surface region. For 5×5, 10×10 and 20×20 cm² field sizes, surface doses with the Markus parallel-plate ion chamber were found 10.81%, 16.61% and 28.06%, respectively. Apipunyasopon et al. [15] investigated the surface dose for 6 MV photon beams using Markus parallel-plate ion chamber. The results were found to be 8.14%, 19.19% and 33.45% at 0 mm for 3×3, 12×12 and 25×25 cm² field sizes, respectively. At the same conditions, Bilge et al. [16] measured the surface doses for 5×5, 10×10 and 20×20 cm² and the results were found to be 10%, 15% and 35%, respectively.

Zhuang et al. [17] measured the surface dose at the effective measurement depth of Markus parallel-plate ion chamber (0.023 mm) for 6 MV photon beams. The result was found 15% . In our study, the surface dose obtained at the same depth was 16.61% . Ying Cao et al. [18] evaluate the superficial dose calculation accuracy of four commonly used algorithms in commercially available TPS by Monte Carlo simulation and film measurements. Surface and buildup doses were measured by the group. In their study, one of the algorithms analyzed for surface dose calculation accuracy is the AAA algorithm which was implanted to Eclipse TPS. According to the study, TPS calculations including AAA algorithm overestimate skin dose near 4.07% when compared by Monte Carlo simulation. It was reported that, the dose distribution calculated by AAA algorithm are in good agreement with MC results within 2–15 mm depth.

Wang et al. [19] studied about the accuracy of skin dose calculations by the AAA algorithm implemented in Varian Eclipse (V.11) system. In order to evaluate the results, The EGSnrc Monte Carlo (MC) simulations were used. Phantom measurements were done by optically stimulated luminescence detectors to validate the accuracy of MC dose calculations. According to their results, AAA v11 underestimates skin doses up to 14% of prescription dose for the patients.

In our study, as seen in Fig. 3 (b), for 10×10 cm² field size, after 1 mm depth, as the results belong to AAA v15.1 and Markus ion chamber are compatible, the dose at 0.07 mm is calculated by AAA v8.9 is lower than both AAA v15.1 and Markus ion chamber. While AAA v15.1 overestimates the skin doses (at 0.07 mm) for all field sizes, AAA v8.9 underestimates.

The low buildup region doses, which we are familiar with in AAA v8.9, is higher than it should be in v15.1. This difference occurred in surface doses after upgrade should not be ignored during the evaluation of radiotherapy plans. It was reported by many groups that TPS cannot calculate surface dose exactly. In radiotherapy, especially if the skin dose of a patient is a critical treatment-limiting parameter, to know the dose calculation behavior of your TPS algorithm in the buildup region is an essential issue.

5. Conclusion

In order to minimize possible calculation errors, algorithms used in treatment planning systems are being developed and upgraded periodically. In clinics, some quality control tests are applied to check the accuracy of the systems after upgrade process. Nevertheless, many quality tests do not include the controlling of the dose calculation in the build-up region.

Algorithms may give different results in surface dose calculations. The different versions of the same TPS algorithms may calculate the surface doses distinctively. After upgrading of TPS algorithms, surface doses should be controlled by different surface dose measurement systems. Before using the new upgraded system, the difference between the calculated surface doses and the measured one must be known correctly.

References

Treatment planning algorithms, Journal of the International Commission on Radiation Units and Measurements, 14(2) (2014), 65–75.

Oelfke and

Scholz , Dose calculation algorithms, New Technologies in Radiation Oncology (2006),

F.M.

Khan , Treatment Planning in Radiation Oncology, 2nd Edition. Lippincott Williams & Wilkins. 2007.

Wang ,

A.J.

Cmelak and

G.X.

Ding , A simple technique to improve calculated skin dose accuracy in a commercial treatment planning system, J Appl Clin Med Phys 19(2) (2018), 191–197.

A.S.

Oinam and

Singh , Verification of IMRT dose calculations using AAA and PBC algorithms in dose buildup regions, J Appl ClinMedPhys 11(4) (2010), 3351.

G.S.

Krishna ,

Srinivas ,

K.M.

Ayyangar and

P.Y.

Reddy , Comparative study of old and new versions of treatment planning system using dose volume histogram indices of clinical plans, J Med Phys 41(3) (2016), 192–197.

B.J.

Gerbi and

F.M.

Khan , Measurement of dose in the buildup region using fixed separation plane-parallel ionization chambers, Medical Physics 17(1) (1990), 17–26.

Oelfke and

Scholz , Dose calculation algorithms, New Technologies in Radiation Oncology (2006), 187–196.

Sievinen ,

Ulmer and

Kaissl , AAA Photon Dose Calculation Model in Eclipse™, Varian Medical Systems, http://www.rtsalon.cn/upload/RTsalon_p_3218_2.pdf, pp. 1–23.

10.

Lu , Dose calculation algorithms in external beam photon radiation therapy, Int J Cancer Ther Oncol 1(2) 2013.

11.

D.E.

Velkley ,

D.J.

Manson ,

J.A.

Purdy and

G.D.

Oliver Jr. , Build up region of megavoltage photon radiation sources, Medical Physics 2(1) (1975), 14–19.

12.

Chan ,

Lee ,

Yu ,

S.P.

Lee ,

Wei ,

Wen ,

Chen ,

Chen and

Tsai , Dosimetric and radiobiological comparison of Cyberknife and Tomotherapy in stereotactic body radiotherapy for localized prostate cancer, JXray Sci Technol 25(3) (2017), 465–477.

13.

Li ,

Xu ,

Lei ,

Zhang ,

Lin and

Li , Evaluation of dose calculations with inhomogeneity correction in intensity-modulated radiation therapy for esophagus cancer, J Xray Sci Technol 26(4) (2018), 657–666.

14.

F.M.

Khan , Treatment Planning in Radiation Oncology, 2nd Edition. Lippincott Williams & Wilkins. 2007.

15.

Apipunyasopon ,

Srisatit and

Phaisangittisakul , An investigation of the depth dose in the build-up region, and surface dose for a 6-MV therapeutic photon beam: Monte Carlo simulation and measurements, J Radiat Res 54(2) (2013), 374–382.

16.

Bilge ,

Cakir ,

Okutan and

Acar , Surface dose measurements with GafChromic EBT film for 6 and 18MV photon beams, Physica Medica 25(2) (2009), 101–104.

17.

A.H.

Zhuang and

A.J.

Olch , Validation of OSLD and a treatment planning system for surface dose determination in IMRT treatments, Med Phys 41 2014.

18.

Cao ,

Yang ,

Qiu ,

Lv , et al., Superficial dose evaluation of four dose calculation algorithms, Radiat Phys Chem 137 (2017), 23–28.

19.

Wang ,

A.J.

Cmelak and

G.X.

Ding , A simple technique to improve calculated skin dose accuracy in a commercial treatment planning system, J Appl Clin Med Phys 19(2) (2018), 191–197.

Phantom	WED	PDD (%)
Depth (mm)	(mm)	5×5 cm²	10×10 cm²	20×20 cm²
0	0,023	10,81	16,61	28,06
0,067	0,07	14,00	19,69	30,87
1	1,068	38,74	43,44	52,48
2	2,113	57,12	60,98	68,19
5	5,248	83,65	85,95	90,00
10	10,473	97,77	98,37	99,24
15	15,698	100,00	100,00	100,00