Dosimetric verification of IMPT using a commercial heterogeneous phantom

Abstract The purpose of this study was to propose a verification method and results of intensity‐modulated proton therapy (IMPT), using a commercially available heterogeneous phantom. We used a simple simulated head and neck and prostate phantom. An ionization chamber and radiochromic film were used for measurements of absolute dose and relative dose distribution. The measured doses were compared with calculated doses using a treatment planning system. We defined the uncertainty of the measurement point of the ionization chamber due to the effective point of the chamber and mechanical setup error as 2 mm and estimated the dose variation base on a 2 mm error. We prepared a HU‐relative stopping power conversion table and fluence correction factor that were specific to the heterogeneous phantom. The fluence correction factor was determined as a function of depth and was obtained from the ratio of the doses in water and in the phantom at the same effective depths. In the simulated prostate plan, composite doses of measurements and calculations agreed within ±1.3% and the maximum local dose differences of each field were 10.0%. Composite doses in the simulated head and neck plan agreed within 4.0% and the maximum local dose difference for each field was 12.0%. The dose difference for each field came within 2% when taking the measurement uncertainty into consideration. In the composite plan, the maximum dose uncertainty was estimated as 4.0% in the simulated prostate plan and 5.8% in the simulated head and neck plan. Film measurements showed good agreement, with more than 92.5% of points passing a gamma value (3%/3 mm). From these results, the heterogeneous phantom should be useful for verification of IMPT by using a phantom‐specific HU‐relative stopping power conversion, fluence correction factor, and dose error estimation due to the effective point of the chamber.


| INTRODUCTION
In recent years, proton beams have become widely used for the treatment of various types of cancer. Active scanning is one of the delivery techniques for proton therapy. 1 The active scanning method moves a spot along the longitudinal and horizontal axis and uses different energy beams to create three-dimensional dose distributions. There are two different delivery methods for scanning proton therapy; the first is single field uniform dose (SFUD) 2 and the second is intensity-modulated proton therapy (IMPT). 3,4 IMPT is more flexible than SFUD and, in general, can deliver a more conformal prescribed dose to the target with lower dose to organs at risks (OARs). IMPT is more sensitive to the uncertainty due to the proton range and mechanical errors than SFUD because IMPT delivers a nonuniform dose distribution to each field. Since the total dose distributions of IMPT are created by summation of some nonuniform dose distributions, it can potentially cause unexpected dramatically hot or cold spots by the combination of inhomogeneity of the human body, machine variations, range uncertainty, and accuracy of the beam modeling or the dose calculation algorithm. 5,6 From this perspective, accurate commissioning, quality assurance (QA) including machine validations and verification of the treatment planning system are fundamental to implementing high-quality IMPT. Heterogeneous phantoms have been effectively used to evaluate the total dose including accuracy of dose calculation and machine variations, and to identify problems that are not revealed by the homogeneous phantom measurements. For this reason, anthropomorphic heterogeneous phantoms are used for the verification of intensity-modulated radiotherapy 7,8 and also used with a thermoluminescent dosimeter (TLD) dosimeter and radiochromic film to verify scattering, pencil beam and uniform scanning plans. 9,10 Furthermore, the results of patient-specific QA of IMPT in some sites using various techniques were excellent. [11][12][13][14][15] However, few studies have focused on the verification for IMPT using commercially available system, especially absolute dose measurement using heterogeneous phantom and ionization chamber. The paper by Taylor et al. 10 reported the summarization for clinical trial credentialing of scattered and scanning proton beam using TLD dosimeter, radiochromic film, several types of IROC's anthropomorphic proton phantoms, and commercial or in-house treatment planning systems (TPSs) that use pencil beam or Monte Carlo-based algorithms. In this study, we report methods and results of verification for the IMPT using a commercially available simple heterogeneous phantom, new TPS, an ionization chamber, and radiochromic film. Our new TPS is the first system that uses the triple Gaussian (TG) model-based pencil beam algorithm. To the best of our knowledge, this is the first report of dosimetric verification of the new TG model-based TPS using commercial heterogeneous phantom and to analyze uncertainties of effective point for the absolute dose measurement using ionization chamber for IMPT.

2.A | Depth and fluence correction for heterogeneous phantom
In this study, we used the RT-3000-New phantom (RT-3000, R-Tech.inc, Tokyo, Japan) that simulates the simple head and neck (HN) and prostate using simulated-bone material. RT-3000 is made from acrylonitrile-butadiene-styrene (ABS) and artificial bone material. Diameters of the simulated-bone material for prostate are 2.5 cm (small) and 4.2 cm (large) and for HN is 1.0 cm, as shown in Fig. 1, 2. In our clinical practice, relative stopping powers (RSPs) were calculated with human tissues as described in ICRU reports 44 16 and 46 17 using the stoichiometric calibration proposed by Schneider et al. 18 However, an error of several percent occurs when clinical computed tomography (CT) values (Hounsfield Units: HU) -RSPs conversion table is applied to the material of RT-3000 phantom. So to use RT-3000 phantom for the verification of IMPT, we created a phantom-specific HU-RSP conversion table using measured RSPs and CT values following the method proposed by Grant et al. 19 We measured RSP of water equivalent materials (soft-tissue phantom) and bone materials (bone phantom) of various thicknesses. A 221.4 MeV proton beam, having a penetration depth of 30.6 cm, was used to measure the change in depth of the distal 90% dose. To create the phantom-specific HU-RSP conversion table, we used the average rate of changes in depth and thicknesses of each material. The fluence correction is another important factor for absolute dose measurement using a solid phantom. 20 In this study, we obtained the fluence correction factor as the ratio of measured dose in the water and that in the soft-tissue phantom material at the same effective depths. Subsequently, an approximate linear formula was derived as the fluence correction factor. The fluence correction for bone phantom is not considered in this paper. The scanned field size was 10 × 10 cm 2 , and the energies and SOBPs used were adapted to each measurement depth. The scanning patterns and measurement depths are summarized in Table 1. To measure at 3 cm depth, we used an energy absorber that had 4 cm water equivalent thickness because the minimum penetration depth of our system was 4 cm. The details on energy absorber will be mentioned in the next section.

2.B | Treatment planning and proton delivery system
The TPS used in this study was VQA (ver 3.0.1 Hitachi Ltd., Tokyo, Japan), a commercially available TPS in Japan. This is the first system that uses the TG model for the dose calculation. The TG model shows distinct difference with other low-dose kernel models in the high energy region where the influence of the secondary particles produced by nuclear interactions in water is increased. In the typical volumetric irradiation, the TG model reported better agreement to the measurement value especially small or large fields and narrow SOBP width condition. 21 The structures, mock clinical target volume (CTV) and OARs of prostate and HN, and dose constraints were adopted from YASUI ET AL.
| 115 AAPM Task Group-119 22 that has produced quantitative confidence limits as baseline expectation values for IMRT commissioning. We used the worst case optimization, 23 the parameters of which were 3 mm for each direction and 3.5% for range uncertainty. The scanning delivery system used in this study was installed in Nagoya Proton Therapy Center. The synchrotron can produce 95 proton beams energies having water penetration distances from 40 to 306 mm. The maximum spot size (1σ) in air at the isocenter plane is 13.8 mm, and the minimum is 4.7 mm. 24,25 To irradiate shallow regions less than 40 mm, we used the energy absorber that had 40 mm water equivalent thickness. The energy absorber can be attached to the beam nozzle and can be used with patient-specific aperture. 26 The spot size of minimum range with energy absorber was 26.7 mm. In this study, we use energy absorber to measure fluence correction factor at 3 cm depth. The energies used in verification plan were from 84.7 to 187.7 MeV for HN planning and from 100.2 MeV to 205.9 MeV for prostate planning. In a clinical setting, we use two parallel-opposed SFUD fields for prostate treatment; however, the simulated prostate plan was created with a 4-fields IMPT plan in this study. The HN plan was created with five fields. and was roughly 1 mm for the PTW 31016 chamber. [26][27][28] The uncertainty of the mechanical isocenter of gantry rotation of the our system has less than 1 mm accuracy. 29 In this report, we have dealt with the error of the measurement point up to 2 mm to the source direction and 1 mm to other directions for each gantry angle, and dose uncertainties were presumed from calculation results of each field dose of TPS. In the simulated prostate plan, the ionization chamber was located at the isocenter, near the rectum region and maximum dose point. In the simulated HN plan, the ionization chamber was located at the isocenter, near to OARs and hot spots. EBT3    Table 2 shows the results of the RSP measurements. HU values showed some dispersion of about 40 HU. In this study, we used RSP values that correspond to each material. The RSPs of soft-tissue phantom were different by less than 2% from the clinical RSP, although the bone phantom that simulated bone was different by more than 30%.

2.C | Experimental measurements
Consequently, the RT-3000 phantom required a phantom-specific HU-RSP conversion table especially for the bone phantom because the phantom was made from artificial materials. The uncertainty of the RSP for the RT-3000 phantom was estimated as 0.45% from the results of multiple measurements of the materials and this uncertainty was sufficiently small. Figure 3 shows the measured results for the fluence correction factor. We adopted a linear approximation formula method to correct measurement values by depth. These results closely resemble those of a previous study 20 that were obtained with a nonmodulated 191 MeV beam, a PMMA, and a Markus-type planeparallel ionization chamber. We assume that the dose distributions were formed by the convolution of nonmodulated proton beams, so the fluence scaling factor of our scanning method was closer to the nonmodulated result of the previous study. However, the T A B L E 1 Scanning patterns and measurement depths to measure fluence correction factor.     | 119 effective measurement point for each field and to estimate the dose variation. The fluence correction by each depth for scanning proton beams is a simple method and results showed good accuracy. Nevertheless, fluence correction factor requires further investigation because that had up to 1.0% variance as shown in Fig. 4. Heterogeneous phantoms can be effective for evaluating the total dose, including accuracy of dose calculation and machine variations, and can help to identify problems that are not revealed by homogeneous phantom measurements.

ACKNOWLEDGMENTS
We thank the members of NPTC, Hitachi. Ltd and R-tech.inc who cooperated with this study.

CONF LICT OF I NTEREST
The authors have no conflicts of interest directly relevant to the content of this article.