Comparing film measurements and Monte Carlo simulations of dose delivered in a pelvic phantom with titanium hip prosthesis

An increasing number of elderly prostate cancer patients with high-density material hip prostheses are referred for external beam Radiotherapy (EBRT). Radiation treatment of pelvis cancer patients with high-density hip prostheses needs special attention due to the artifacts created in the computed tomography (CT) field of view and the radiotherapy dosimetry challenges. This study investigated the pelvic prostate point dose with and without titanium hip prosthesis using a 0.6 cc PTW Farmer ionization chamber, EBT3 Gafchromic films, compared with the EGSnrc Monte Carlo (MC) simulation dose distribution. The doses were measured and simulated in a locally made pelvic phantom. MC and measured doses were compared with the Treatment Planning System (TPS) calculated prostate point dose. The ionization chamber, EBT3 Gafchromic films, and MC doses have a maximum deviation of 6.3 %, 5.7 %, and 7.4 % for 6 MV and 4.2 %, 4.7 %, and 5.5 % for 15 MV photon beam, respectively, when compared with TPS calculated dose. There is a significant difference between the prostate point dose measured with ionization chamber, EBT3 Gafchromic film in comparison MC simulated doses. The MC simulation dose shows the highest deviation especially on the lateral field passing through the prosthesis.


Introduction
Prostate cancer develops mostly in older men who may have osteoarthritis [1] and who may require a hip prosthesis.High-density prostheses manufactured using materials such as titanium or steel cause significant perturbation of a radiation beam and create image artifacts in the CT scanner.Artifacts are a common phenomenon in X-ray CT [2].In CT scan, the attenuation coefficients of high-density materials are much higher than those of soft tissues because the X-ray beam is highly attenuated by metals and the insufficient number of photons reaches the detector resulting in corrupted (noisy) projection data [3,5].
The image artifacts created by high-density prostheses in the field of view of a CT scanner are due to four main factors, namely: beam hardening [4,5], scatter, noise [2,6] and photon starvation [4].Beam hardening occurs because low-energy photons are attenuated to a greater degree than high-energy photons when passing through high-density scanned materials [4,5].Poisson noise is caused by statistical errors of low photon counts which result in random, thin, bright, and dark streaks that appear in the direction of greatest attenuation [8].Photon starvation artifacts are caused when photons transverse a high-attenuation coefficient material [5,6].
These factors degrade the capabilities of CT images to provide correct information about the HU and electron density of structures.These artifacts conceal the true anatomy of organs and may render organ borders indistinct [6].The CT data sets with artficats are used in treatment planning system to delineate the tumor and critical organs for treatment planning.The CT structures affected by these artifacts during prostate cancer treatment planning are the rectum, bladder, femoral head, and prostate.
CT data set with scatter artifacts arising from high-density metal prosthesis may require dosimetrist and physicist involvement to correct.This can be done by contouring the metal prosthesis and scatter artifacts, and manually assigning a CT number to those areas to correspond to the desired relative electron density (RED) [7].Several methods have been proposed for the reduction of metal artifacts in CT images.Filter back projection (FBP) is the standard reconstruction method for most scanners.The projection data are filtered to sharpen edges and the filtered data are then back projected [8].Interpolation-based methods are also used to correct the corrupted projection data by using the uncorrupted data on both sides of the metal trace.However, reducing metal artifacts using this method introduces new artifacts and can increase the obscurity of the normal tissue structures [9].
Metal artifacts reduction (MAR) algorithms compensate for photon starvation caused by attenuating material in the X-ray path [10].However, most MAR algorithms require faster computer chips [8].Metal Artifact Reduction in Image space (MARIS TM ) (Siemens Healthcare GmbH, Erlangen, Germany) was introduced to improve image quality for radiation therapy planning [2].Metal artifact reduction for orthopedic implants (O-MAR) (Philips Healthcare, Cleveland, Oh, USA) is an iterative projection modification method optimized for imaging orthopedic devices [5].However, there is a concern that MAR methods may not provide accurate attenuation values near the high-density material and may hinder the ability to detect lesions by smoothing interfaces in an attempt to reduce noise [11].
Ali, [4] indicated that increasing the tube peak voltage can also be used to reduce metal artifacts but this solution is insufficient in terms of image quality for diagnosis.Increasing the tube peak voltage reduces the metallic artifact to a minor degree but may not improve the scatter caused by imaging hip prostheses.Increasing the tube current also increases the radiation dose to the patient [5].
Successful radiotherapy depends on the accuracy of the dose delivered to the target volume while the dose to organs at risk is kept as low as possible [12].The accuracy of treatment planning dose calculation algorithms determines the accuracy of the dose delivered to the patient [13].Most of these algorithms have been designed for lower densities and need to accurately model the absorption of the scattering properties of high-density metals [14].
It has become a common practice in radiotherapy treatment planning to avoid directing radiotherapy treatment fields through a high-density hip prosthesis [14] because of the higher energy absorption of the prosthesis compared to normal tissue and thus the poor penetration of the beam through the prosthesis.Reft et al., [15] stated that although this method avoids the dosimetric complications of the high-density metallic inhomogeneity, it becomes challenging to achieve an acceptable dose distribution coverage of the target (PTV/Prostate) whilst attempting to minimize the dose to the critical organs (e.g., bladder and rectum).Some treatment beams may exit through the prosthesis to obtain a homogenous dose distribution.
Monte Carlo (MC) calculation methods have been used in radiotherapy for modeling of electron, photon and proton beams as well as dose distributions for many years.Studies have shown the potential of MC simulation to accurately calculate high-density scatter radiation dose.For this study, EGSnrc MC code was utilized to calculate dose distribution in a locally made pelvic phantom CT.
The aim of the study is to optimize the dose distribution of prostate 3D conformal treatment in the locally made metallic hip prosthesis phantom.This was achieved by investigating the pelvic prostate dose distribution with and without a hip prosthesis using 0.6 cc PTW Farmer ionization chamber, EBT3 Gafchromic films in comparison with the Monte Carlo simulated results.

Methods and materials
The description of the materials and methods are divided into 3 sections, namely, (1) locally made pelvic phantom design, (2) experimental measurements, and (3) Monte Carlo simulation techniques.The locally made pelvic phantom was employed to verify Treatment Planning System (TPS) calculated dose on a linear accelerator using an 0.6 cc PTW farmer ionization chamber and EBT3 Gafchromic films incorporated with doses simulated using EGSnrc MC code.

Locally made pelvic phantom design
The locally made pelvic phantom was constructed from superflab gel bolus slices, Nylon-12, dental wax, bone material, and a titanium hip prosthesis.The radiological property of superflab gel bolus and dental wax was investigated using two methods: Firstly by measuring the linear attenuation coefficient.The waterequivalence of superflab gel bolus, and dental wax was investigated by measuring central axis (CAX) transmission (the linear attenuation coefficient) and calculating mass linear coefficient.The linear attenuation of superflab bolus and dental wax was obtained by measuring the transmission factors through different thicknesses of superflab gel bolus and dental wax.These measurements were compared to those taken with an RW3 Solid Water Phantom for a similar setup.Secondly via relative density obtained for CT number directly from a CT scan.The dental wax was melted and poured directly into a mould to minimize air gaps between the bone material, Nylon-12, and the hip prosthesis as they are not flat.The mould was then sandwiched between the super flab gel boluses as shown in figure 1. Nylon-12 was used to replicate the prostate with a cavity to fit the ionization chamber and film insert in its centre.The locally made pelvic phantom is 30 cm in height, 30 cm in width and 17 cm thick.(30 x 30 x 17 cm 3 ).

Experimental measurements
The locally made pelvic phantom was scanned on a 12-bit depth images Philips CT Scanner and the CT images were imported into a Varian Eclipse 16.1 treatment planning system.The organs of interest were delineated, and no artifact density corrections were performed.Four different plans for 6 MV and 15 MV photon beams were created consisting of anterior (gantry 0 0 ) and posterior (180 0 ) fields and left (90 0 ) and right (270 0 ) lateral fields.One lateral field passed through the bone of the normal hip and one through the hip prosthesis.The plans were prescribed to 2 Gy x 10 fractions and optimized to ensure that the prescribed dose covered at least 100 % of the dose of the ionization chamber/film insert.
The phantom was positioned on a Varian Clinac IX linear accelerator as scanned and a cone beam CT image was acquired to verify the setup and the position of the ionization chamber and films inside the phantom as shown in figure 2. The plan doses were delivered on the phantom and the prostate point dose was measured using 0.6 cc PTW Farmer ionization chamber.The ionization chamber was calibrated at the National Metrology Institute of South Africa (NMISA).A plastic rod with EBT3 Gafchromic films was also inserted inside chamber (cavity) of the phantom.Before using EBT3 film for the prostate point dose measurement, the dose calibration curve was obtained by irradiating small pieces of film using doses ranging from 0.25 Gy to 6 Gy.Film calibrations allow the conversion of optical density to dose for future measurements.The optical density can be related to the dose by applying a fitted function known as a calibration curve.
After irradiation, the films were scanned, and the pixel values were converted into doses using EBT3 film calibration curve.The doses measured with the ionization chamber and with EBT3 films were compared with the prostate point dose from the TPS plans.

EGSnrc Monte Carlo Simulation
The BEAMnrc tool of the EGSnrc MC package (National Research Council, Canada) was used to model a 6 MV and 15 MV Varian Clinac IX beam.The accuracy of the Clinac model was validated by comparing the MC simulated dosimetric characteristics (percentage depth dose and profiles) with the commissioning data measured in water.Consequently, the pelvis phantom CT images were converted from dicom (.dcm) format to DOSXYZnrc readable file using the CTcreate tool.The phantom was simulated for different fields: lateral fields were simulated using 6 x 6 cm 2 and anterior/posterior using 10 x 10 cm 2 for both 6 MV and 15 MV photon beams.The resultant 3ddose files from DOSXYZnrc were analyzed using the Voxel Interactive Contour Tool for Online Radiation Intensity Analytics (VICTORIA) program.

Results and Discussion
The dental wax, superflab bolus, and Nylon-12 each have a density close to that of water (1 g/cm 3 ).The calculated linear attenuation coefficient of the RW3, dental wax, and superflab gel bolus were found to be 0.046 cm -1 , 0.043 cm -1, and 0.045 cm -1 .The mass attenuation coefficients of the RW3, dental wax, and superflab gel bolus were 0.0457 cm 2 /g, 0.0449 cm 2 /g, and 0.0451 cm 2 /g, respectively.The deviation of mass attenuation coefficient was found to be 1.3 % and 1.7 % for superflab gel bolus and dental wax, respectively, when compared to the RW3 slab mass attenuation coefficient.
Ade et al., [16] reported the RW3 slab linear attenuation coefficient of 0.0486 cm -1 and the mass attenuation of 0.0465 cm 2 /g measured for 6 MV photon beam.These materials (dental wax and superflab bolus) were chosen as they are regularly used in radiotherapy to compensate for missing tissue.The linear and mass attenuation coefficient of Nylon-12 were not done due to having access to a small piece (prostate piece).However, in the literature, a difference of 1.5 % was found between the mass attenuation coefficient of Nylon-12 (0.0458 cm 2 /g) and RW3 slabs (0.0465 cm 2 /g) [16].
The validation of EGSnrc MC simulated dose distribution agreed well with commissioning data within 2 %.Only the percentage dose depth for different field sizes at Dmax results are shown in table 1. Validating the linear accelerator gives us the opportunity to use an accurate model for further dosimetric studies even in a complicated environment where physical measurement is unattainable.The prostate point doses measured with EBT3 Gafchromic films and ionization chamber shows good agreement with the Eclipse TPS calculated dose compared to MC simulated doses for all plans.However, the lateral fields passing through the prosthesis reveal that the ionization chamber, EBT3 Gafchromic films and MC dose have a maximum deviation of 6.3 %, 5.7 % and 7.4 % for 6 MV and 4.2 %, 4.7 % and 5.5 % for 15 MV photon beam, respectively as shown in table 2 and 3.The MC dose distribution of the field passing through the prosthesis is shown in figure 3.   [17] used two different in-house phantoms to evaluate the treatment planning dose calculation percentage error of a static 6 MV photon beam passing through the metal hip prosthesis.First method involves dental wax wrapped around the metal hip prosthesis and placed on top of a PTW 2D array, while the second method involves titanium insert was placed inside a slab phantom and measured using film and Thermoluminescent dosimeter (TLD).The results showed the percentage difference on the surface of the implant was to be 1.6 % for the 2D array, 14 % for films and 4 % for TLD.
Numerous studies have also employed EGSnrc MC simulation to evaluate the effect of hip prosthesis and compared to the TPS algorithms.Jayamani et al., [18] investigated the effect of titanium rod on 12-bit, 12bit extended and 16-bit depth CT data set for 6 MV photon beam using EGSnrc MC simulation and compared with Monaco TPS calculated dose.In their study, a wax phantom with a titanium rod was used to evaluate the dose at Dmax, at the entrance point before the titanium rod, at the point within the titanium and the exit point from the titanium rod.The Dmax agreement between the TPS calculated dose and MC calculated dose was found to be 1.31 %, 1. 24 % and 1.35 % and for the entrance point before titanium was found to be 18.65 %, 14.47% and 18.56 % for the 12-bit, 12-bit extended and 16-bit depth respectively.The agreement within the titanium rod was found to be 12.72 %, 6.80 % and 4.95 % and for the exit point from the titanium rod was found to be 0.72 %, 13.68 % and 6.33 % for the 12-bit, 12-bit extended and 16-bit depth, respectively.

Conclusion
The treatment planning system overestimated the prostate point dose as evidenced from the deviations reported in table 2 and 3.The maximum deviation calculated in this study for EBT3 Gafchromic films, ionization chamber and MC simulation comes from the fields passing through the prosthesis for both 6 MV and 15 MV photon beams with MC simulated dose showing the maximum deviation.

Figure 1 .
Figure 1.Pictorial section of phantom showing 5 cm thick moulded dental wax on top of 6 cm superflab gel bolus.

Figure 2 .
Figure 2. CBCT position verification image with the chamber inside the prostate (in red) cavity.

Figure 3 .
Figure 3. MC simulated 6 MV photon beam dose distribution of the left lateral field (Gantry 90 0 ) passing through prosthesis.

Table 1 .
The percentage depth dose (PDD) comparisons between MC simulated and commissioning data at Dmax for 6 MV and 15 MV photon beam for different field sizes.

Table 2 .
Comparison of Ionization chamber, EBT3 film measurements with MC simulated prostate point dose for 6 MV photon beam and the percentage difference compared to TPS calculated point dose indicated.6

Table 3 .
Comparison of Ionization chamber, EBT3 Gafchromic film measurements with MC simulated prostate point dose for 15 MV photon beam and the percentage difference compared to TPS calculated point dose indicated.