Dose comparison between Gafchromic film, XiO, and Monaco treatment planning systems in a novel pelvic phantom that contains a titanium hip prosthesis

Abstract The presence of metallic prostheses during external beam radiotherapy of malignancies in the pelvic region has the potential to strongly influence the dose distribution to the target and to tissue surrounded by the prostheses. This study systematically investigates the perturbation effects of unilateral titanium prosthesis on 6 and 15 MV photon beam dose distributions using Gafchromic EBT2 film measurements in a novel pelvic phantom made out of a stack of nylon slices. Comparisons were also made between the film data and dose calculations made on XiO and Monaco treatment planning systems. The collapsed cone algorithm was chosen for the XiO and the Monte Carlo algorithm used on Monaco is XVMC. Transmission measurements were taken using a narrow‐beam geometry to determine the mass attenuation coefficient of nylon = 0.0458 cm2/g and for a water‐equivalent RW3 phantom, it was 0.0465 cm2/g. The perturbation effects of the prosthesis on dose distributions were investigated by measuring and comparing dose maps and profiles. The magnitude of dose perturbations was quantified by calculating dose enhancement and reduction factors using field sizes of 3 × 3, 5 × 5, 10 × 10, and 15 × 15 cm2. For the studied beams and field sizes, dose enhancements between 21 and 30% and dose reductions between 15 and 21% were observed at the nylon‐prosthesis interface on the proximal and distal sides of the prosthesis for film measurements. The dose escalation increases with beam energy, and the dose reduction due to attenuation decreases with increasing beam energy when compared to unattenuated beam data. A comparison of film and XiO depth doses for the studied fields gave relative errors between 1.1 and 23.2% at the proximal and distal interfaces of the Ti prosthesis. Also, relative errors < 4.0% were obtained between film and Monaco dose data outside the prosthesis for 6 and 15 MV lateral opposing fields.

in a novel pelvic phantom made out of a stack of nylon slices. Comparisons were also made between the film data and dose calculations made on XiO and Monaco treatment planning systems. The collapsed cone algorithm was chosen for the XiO and the Monte Carlo algorithm used on Monaco is XVMC. Transmission measurements were taken using a narrow-beam geometry to determine the mass attenuation coefficient of nylon = 0.0458 cm 2 /g and for a water-equivalent RW3 phantom, it was 0.0465 cm 2 /g. The perturbation effects of the prosthesis on dose distributions were investigated by measuring and comparing dose maps and profiles. The magnitude of dose perturbations was quantified by calculating dose enhancement and reduction factors using field sizes of 3 9 3, 5 9 5, 10 9 10, and 15 9 15 cm 2 .
For the studied beams and field sizes, dose enhancements between 21 and 30% and dose reductions between 15 and 21% were observed at the nylon-prosthesis interface on the proximal and distal sides of the prosthesis for film measurements.
The dose escalation increases with beam energy, and the dose reduction due to attenuation decreases with increasing beam energy when compared to unattenuated beam data. A comparison of film and XiO depth doses for the studied fields gave relative errors between 1.1 and 23.2% at the proximal and distal interfaces of the Ti prosthesis. Also, relative errors < 4.0% were obtained between film and Monaco dose data outside the prosthesis for 6 and 15 MV lateral opposing fields.

| INTRODUCTION
A common treatment modality for prostate cancer is external beam radiation therapy. It is aimed at delivering a lethal radiation dose to malignant tissues so as to provide a high probability of tumor control while sparing or inducing minimal damage to adjacent normal tissues. Therefore, radiotherapy is directed toward keeping normal tissue complications within acceptable limits while attaining a high therapeutic ratio. However, an increasing number of patients requiring megavoltage photon beam radiotherapy for malignancies in the pelvic or hip region have metal implants or prostheses which could shadow the target and influence the dose distribution leading to a dramatic difference in clinical outcome. [1][2][3][4] Implants vary in size, composition, and geometry, and the choice of an implant material depends on factors such as corrosion, fatigue resistance, and mechanical strength. 5 Commonly used metals and alloys for implants include stainless steel, Co-Cr-Mo, and Ti. 3,5,6 Carolan et al. pointed out that Co-Cr-Mo alloy with a high relative electron density (6.79-6.9) is likely to have a greater impact on dose distribution than steel (6.55-6.61) and Ti (3.72-3.76) which have low electron densities. 3 Mesbahi and Nejad, however, observed that the attenuation effect of prostheses is density dependent with steel (q = 8.1 g/cm 3 ) showing the greatest impact followed by Co-Cr-Mo (q = 7.8 g/cm 3 ) and Ti (q = 4.54 g/cm 3 ) showing the least effect. 6 The majority of hip prostheses are composed of Co-Cr alloys as they are considered to have the best combination of corrosion, fatigue resistance, and mechanical strength. 5 The high Z and high density of metallic prostheses relative to water yield challenges for radiotherapy dose computation when beams pass through these devices because the dose attenuation through a prosthetic device during pelvic irradiation could be significant. 2,[5][6][7][8][9][10] In addition, the drastic changes in electron scattering characteristics near interfaces due to sudden and extreme changes in density are also challenging for most dose calculation algorithms that are not Monte Carlo (MC) or Acuros as MC and Acuros algorithms can account for the effects of heterogeneities in patient dose calculation.
It is known that there is a decrease in tumor control due to reduced target dose from beam attenuation of the prosthesis 3 or an increase in complication rates due to the local dose perturbations caused by prosthetic implants. 1 Perturbations of absorbed dose distribution occur as a result of the increased attenuation of the radiation beam by the prosthetic device and the changes in electron scatter or photon interactions (photoelectric effect and pair production) that occur at the bone-metal interface. Even though oblique beams are usually chosen to minimize or avoid the shadowing effect of the prostheses, this cannot always be accomplished. 4 It could also cause an increased dose to adjacent structures such as the rectum. 3,11 A survey of 30 institutions conducted by the AAPM TG-63 indicated that the number of patients with prostheses, which could affect their radiation therapy, was 1-4% of the total number of patients. 1 The survey also indicated that there was no general agreement on how to manage the treatment for patients having prostheses. Some institutions ignore their presence, while others try to modify the beam orientation to avoid the prostheses even if extra dose is delivered to adjacent critical structures. With an increasing aging population, it is expected that the number of patients with prostheses is likely to increase due to conditions such as osteoarthritis and dysfunctional hip joints which may require hip replacement.
It is understood that the scientific understanding and approach of clinical dosimetry for the presence of metallic prostheses during irradiation of pelvic malignancies is still a challenge. 2,9,10 Also, the dose perturbation due to these prostheses could affect clinical outcome due to its significance and so it cannot be ignored. 7,8 As a result, it is necessary to expand the information available in literature with current data which is the motivation of the present study.
A number of researchers have attempted to quantify the dose perturbations due to prostheses by performing dose measurements either in phantom (usually liquid or solid water phantoms) containing the prosthesis or by computing with a treatment planning system (TPS) or Monte Carlo (MC) methods. [2][3][4]6,[8][9][10][11][12][13][14] Biggs and Russell used ionization chamber dosimetry systems and a water tank to measure the effects of a hollow femoral head prosthesis on the dose from lateral fields to the pelvis for megavoltage photon beams. 11 Sibata et al. 5 measured dose in water with film and an ion chamber to evaluate prosthesis-induced attenuation for 6 and 18 MV photons. The prosthetic models had varying size and composition of Co-Cr, Ti, and stainless steel. Some authors used diodes to measure the dose attenuation for a 10 9 10 cm 2 6 MV photon field due to the presence of a Co-Cr-Mo hip prosthesis in a water tank. 3 Others used film to measure the dose attenuation along the length of a Ti alloy hip prosthesis for 6 and 15 MV photon beams. 8 Spezi et al. evaluated the dosimetric characteristics of commonly used prosthetic implant materials using 6 and 18 MV photon beams. The materials were cut into cylinders and measurements were conducted using ion chambers in water and RW3 solid water phantoms. The measured data were compared with calculations based on Monte Carlo treatment planning models. 2 Kung et al. investigated the feasibility of using IMRT for treating patients with metallic prostheses. 12 Erlanson and Franz en measured the dose distribution effects caused by a hip prosthesis with small silicon diodes for 6, 20, and 50 MV photons when treating pelvic cancer. 14 Others studied a Ti alloy prosthesis in a water phantom using MC simulations and a TPS calculations to study the perturbations due to metallic implants for 6 and 18 MV photon beams. 4 As reported by the researchers, the degree of the dose perturbations varied between 2 and 64%. However, the usual use of water phantoms and the lack of much detail are limitations of most the studies. A more convenient approach will be the use of a more realistic tissue-mimicking medium such as a water-equivalent solid pelvic phantom for dose perturbation measurements for a meticulous and systematic study.
In this paper, the dose perturbation effect of Ti prosthesis for 6 and 15 MV photon beams was meticulously studied in a novel realistic pelvic phantom consisting of a stack of nylon slices with bone and Ti embedded in each layer to form a unilateral Ti prosthesis.
Dose measurements were made with Gafchromic film to determine dose perturbation factors for a range of field sizes. Single-and bilateral-field film studies were also compared with dose calculations ADE AND PLESSIS | 163 using a CMS XiO and Monaco TPSs. To the best of our knowledge, reports for dose perturbation data for a Ti prosthesis evaluated in such a pelvic phantom for 6 and 15 MV photon beams are scarce.
Also, a comparison of Monaco TPS against film measurements for a study involving Ti prosthesis has not been reported previously.

2.A | The pelvic phantom
The phantom is locally developed with a built-in Ti hip prosthesis and was designed for film dosimetry as shown in Fig. 1 and filler materials which comprise silicon dioxide (25.5) and calcium carbonate (23.5). 15 As the phantom is designed for film dosimetry, removable inter-slice plastic plate templates were manufactured to allow precise film cutting that could fit between the slices for measurements. The design of the phantom is such that air gaps between the nylon slices which could influence dose measurements are minimized. This is further achieved by using clamps to fasten the nylon slices so as to keep the phantom airtight during measurements. The diameter of the Ti disk in the plane of measurement considered in this study is 2.7 cm [ Fig. 1 (b)] and the width of the bone material on the opposite side is about 4.5 cm. There is a thin layer of tissue material (nylon) between the prosthesis and bone, which forms the bone-prosthesis interface.

2.A.1 | Nylon water-equivalence
The water-equivalence of Nylon-12 was established by measuring central axis (CAX) transmission data through slabs of this material as well as water-equivalent RW3 slabs. 16,17 Measurements were made in narrow-beam geometry using a 6 MV photon beam and a 0.6 cc Farmer-type ion chamber connected to a PTW UNIDOS E electrometer. The chamber was housed inside a block of Perspex with 0.8 cm buildup. The block, containing the chamber, was placed at an SSD of 200 cm and 6 MV transmission measurements were taken with the phantom slabs placed at an SSD of 100 cm for a set field size of 2 9 2 cm 2 defined at 100 cm SSD (Fig. 2). 300 monitor units (MU) were set up in each measurement to ensure high signal-tonoise ratio. Transmission measurements (R x ) were made with different thicknesses of attenuating material for Nylon-12 and RW3, ranging from 1 to 10 cm in steps of 1 cm.
From these measurements and the application of exponential attenuation law (1), the linear attenuation coefficient (l) was calculated, as well as the mass attenuation coefficient (l m = l/q). 18,19 A density value of 1.01 g/cm 3 was used for Nylon-12.
In eq. (1), R x is the ionization signal transmitted through the attenuating phantom material of thickness x (cm), R o is the initial open beam signal, and T x is the transmission factor. The linear attenuation coefficient was then determined from a least square fit of an exponential function through the transmission data points on a T x vs x graph.
As electron scatter characteristics are most significant near water and bone-prosthesis interfaces as transient charge particle equilibrium is disturbed at tissue/bone-prosthesis interfaces, another dosimetric parameter to consider for the water-equivalence of nylon is to evaluate its electron stopping power characteristics relative to water. Table 4 of the AAPM TG-21 provides a list of ratios of average, restricted stopping powers of medium to air for some materials including nylon for photon spectra ranging from 2 to 45 MV. 20 The discrepancy between the values of water and nylon are within 1% implying that nylon and water have similar stopping power characteristics.

2.B | Film calibration and phantom measurement
All dose measurements reported in this study are from exposures of XðDÞ ¼ a þ ½b=ðD À cÞ in its center. 23 This avoided OD measurement artifacts near film edges. 24 The process was repeated for all film pieces. An Epson Perfection V330 Photo flat-bed document scanner with a resolution of 72 dots per inch (dpi) was employed to read the films. Film images were scanned as raw 48-bit RGB (16 bits per color) and saved in tagged-image-file format (TIFF) similar to procedures reported in literature. 23,25,26 These images were processed using information in the 16 bit red (R) channel of the RGB tiff images.
The delivered dose D versus measured OD was then fitted employing the analytical function depicted in (2). The OD was determined from the pixel reading in similar procedures as reported elsewhere. 23 2.C | Dose distribution EBT2 film measurements Beam setups/directions for dose perturbation measurements. A 10 9 10 cm 2 field was used for lateral beams and field sizes of 3 9 3, 5 9 5, 10 9 10, and 15 9 15 cm 2 were used for AP beams. were computed as follows.

3.A | Transmission measurements
Transmission factors for different thicknesses of nylon and waterequivalent RW3 materials are shown in Fig. 4. Table 1  T A B L E 1 Linear and mass attenuation coefficients of Nylon-12 and water-equivalent RW3 plastic materials determined using a 6 MV photon beam.
Phantom material l (cm À1 ) l m (=(l/q)) (cm 2 /g)  The DPF values for the four fields are shown in Table 2 Table 3 shows DPF values in the proximal region of the Ti prosthesis in relation with distance from its interface. From Table 3  further away from the surface regardless of energy, and (c) at any given distance from the prosthesis, the sensitivity of the DPF to field size variation is found to be greater for lower energies. The narrow distance window through which the proximal DPF is observed suggests a low range of backscatter electrons which in turn suggests low-energy backscatter electrons. 28 The range of backscattered electrons is also dependent on the photon energy. Figure 10 shows the average values of the DPF for all four-field sizes as a function of distance from the proximal interface. It is observed that the DPF due to backscatter is higher for 15 MV compared to 6 MV.  Table 4. The relative errors between the film and XiO TPS dose data calculated at the proximal and distal interfaces of the Ti Prosthesis are presented in Table 4. For the studied fields and beams, relative errors between 1.9 and 23.2% were obtained at the  (Table 5), but with considerable variation from one study to another. The magnitude of the perturbations varies between 2 and 64% depending on the size, thickness, mass density, design, and composition of the prosthesis as well as the differences in multiple scatter of the secondary electrons and the incident beam energy. 1,2,4,[8][9][10][11][12][13][14] Additionally, most data are limited to the perturbation effect due to photon attenuation behind an T A B L E 3 Variation of the DPF with distance from the tissue-prosthesis interface on the proximal side of the prosthesis. The uncertainty in the data is 2%.  inhomogeneity usually placed in normal water or plastic phantoms.
In this study, a novel pelvic phantom that simulates patient geometry with a built-in Ti prosthesis is employed. Dose perturbations were systematically investigated along depth dose curves and not just at the beam entry or beam exit of the prosthesis as often reported in literature. Presented in Table 5  In this study, the following observations can be made from It has, however, been shown that for lower Z materials such as bone and aluminum, the BSDF is roughly constant with photon energy up to 10 MV and then falls off at higher energies. 28 For higher Z materials such as lead, the BSDF increased from 60 Co and peaked at 10 MV. 28 Little variation of the dose enhancement with field size has been reported. 11,28 For various materials in photon beams from 6 to 24 MV, the BSDF was found to be constant with field size between 4 9 4 and 20 9 20 cm 2 , except for lead where the BSDF was lower at smaller fields and saturates at 8 9 8 cm 2 . 28 This field size independence of the BSDF was attributed to electron transport rather than photon backscattering. 28 The findings of the present study (Table 2), however, show that the dose enhancement at the proximal interface of the prosthesis increases with decrease in