Feasibility of a 3D-printed anthropomorphic patient-specific head phantom for patient-specific quality assurance of intensity-modulated radiotherapy

This study evaluated the feasibility of utilizing a 3D-printed anthropomorphic patient-specific head phantom for patient-specific quality assurance (QA) in intensity-modulated radiotherapy (IMRT). Contoured left and right head phantoms were converted from DICOM to STL format. Fused deposition modeling (FDM) was used to construct an anthropomorphic patient-specific head phantom with a 3D printer. An established QA technique and the patient-specific head phantom were used to compare the calculated and measured doses. When the established technique was used to compare the calculated and measured doses, the gamma passing rate for γ ≤ 1 was 97.28%, while the gamma failure rate for γ > 1 was 2.72%. When the 3D-printed patient-specific head phantom was used, the gamma passing rate for γ ≤ 1 was 95.97%, and the gamma failure rate for γ > 1 was 4.03%. The 3D printed patient-specific head phantom was concluded to be highly feasible for patient-specific QA prior to complicated radiotherapy procedures such as IMRT.


Introduction
The effectiveness of radiotherapy is maximized by delivering the highest possible dose of radiation to a tumor while minimizing the effect on normal tissues [1,2]. In particular, treatment techniques such as intensity-modulated radiotherapy (IMRT) and volumetric-modulated arc therapy (VMAT) optimize the composite dose distribution and deliver a non-uniform fluence from any given position of the treatment beam to the patient [3]. The dose distribution needs to be calculated with a radiation treatment planning system (RTPS), and patient-specific quality assurance (QA) is needed to validate the dose distribution. Dose distributions can be measured by either using a chamber (i.e., point dosimetry) or a film or diode array detector (i.e., two-dimensional dosimetry) [4,5]. For patient-specific QA using a film or array detector, the gamma index is currently used extensively to validate the difference between the calculated and measured doses [6,7]. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 In recent years, various cases of radiotherapy using 3D printing technology have been reported. For instance, Ehler et al. [8] constructed the head and neck components of the Rando phantom (The Phantom Laboratory, Salem NY) with a 3D printer. Park et al. [9] created a 3D customized bolus and used it as a compensator in IMRT of a patient with Kimura's disease involving the auricle. Choi et al. [10] utilized a 3D printer to evaluate the effectiveness and accuracy of a patient-specific bolus in electron beam therapy, while Kim et al. [11] created a customized bolus with a 3D printer to analyze its effectiveness in comparison with the conventional Superflab bolus in photon beam radiotherapy. For proton therapy, Ju et al. [12] concluded that the physical accuracy and dosimetric characteristics of a proton range compensator (RC) created with a 3D printer and the conventional RC constructed with a computerized milling machine are comparable. Our institution [13] recently fabricated a customized 3D bolus for an irregular surface by using a 3D scanner and 3D printer.
Ehler et al. [8] pointed out that a QA method based on using a 3D phantom constructed with a 3D printer provides a better end-to-end test than traditional per-patient quality control tests because it matches the clinical patient workflow.
To realize the advantages of this patient-specific phantom using a 3D printer, this study evaluated the feasibility of using a 3D-printed anthropomorphic patient-specific head phantom of a head and neck cancer patient for patient-specific QA in IMRT.

Materials and methods
Construction procedure for a patient-specific head phantom The computed tomography (CT) image used in this study was a Digital Imaging and Communications in Medicine (DICOM) file intended for use in the radiotherapy treatment plan for head and neck cancer patients. The CT image was scanned at 3 mm intervals with a Philips Big Bore Brilliance CT Scanner (Philips Medical, Eindhoven, Netherlands). The scanned CT image was then sent to an Eclipse 8.6 RTPS (Varian Medical System Inc., Palo Alto, CA). The image was scanned in order to increase the smoothing level of the 3D-printed head phantom and reconstructed in intervals of 0.5 mm at the Eclipse RTPS because of its thickness. The "search body" function in the contouring tab of the Eclipse system was used to contour the body automatically. Just like that used clinically for patient CT, the lower threshold was set to −300 Hounsfield units (HU). The contoured body structure was partitioned to the left and right to measure the radiation dose, as shown in An anthropomorphic patient-specific head phantom was constructed by using the fused deposition modeling (FDM) method with a Dimension 1200 Series SST printer (Stratasys, Eden Prairie, MN). ABSplus (Stratasys, Eden Prairie, MN) was chosen as the material to construct the head phantom. CT scan using a patient-specific head phantom and treatment plan In order to obtain the CT DICOM for the radiotherapy treatment plan using the patient-specific head phantom in Fig 2(A), the image was scanned at a thickness of 1 mm. For other conditions, the same protocol was used for the head phantom as that for a CT of a patient's head. The patient-specific head phantom was fixed into position with a DUON TM mask (Orfit The treatment plan for the patient-specific head phantom was formulated by performing fixed-gantry IMRT at the angles 0˚, 40˚, 80˚, 120˚, 160˚, 200˚, 240˚, 280˚, and 320˚. In this treatment plan, the same tolerance and priority values used for a real patient's target and normal organs were applied and re-optimized. Fig 3(F) shows the dose distribution calculated for the 3D-printed patient-specific head phantom in the sagittal plane. The treatment plan re-optimized for the head phantom was also transferred to I'mRT MatriXX (IBA Dosimetry GmbH) for calculation. Fig 3(E) shows the dose distributions calculated with I'mRT MatriXX in the frontal plane. The ion chambers were arranged in a 2D array on the frontal plane, which made it possible to compare the measured and calculated doses. The sagittal plane of the patient- Evaluation of the dose distributions with I'mRT MatriXX and a patientspecific head phantom: gamma index (γ) To verify the calculated and measured dose distributions, we used I'mRT MatriXX with the Plastic Water 1 phantom MULTICube (IBA Dosimetry GmbH) and the 3D-printed patient specific head phantom. In recent years, the gamma index has been widely used to compare calculated and measured doses [6,7,14].
Janusz et al. [14] defined the gamma index (γ) to compare the distributions of the calculated and measured doses as follows: where rðr m ; rÞ ¼ br À r m c ð2Þ and dðr m ; is the dose difference (DD) at the position r m . Here, r m is a single measurement point, r c is the spatial location of the calculated distribution relative to the measured point, Δd M is the passing criterion for the distance-to-agreement (DTA), and ΔD M is the passing criterion for the DD. The calculation passes if γ 1 and fails if γ > 1. In order to compare the calculated and measured doses for gamma evaluation, the acceptable DD and DTA values were set to 3% and 3 mm, respectively. Patient-specific quality assurance with I'mRT MatriXX and a patientspecific head phantom I'mRT Matrixx. A Novalis Tx linear accelerator system (Varian Medical Systems, CA, USA and BrainLAB, Feldkirchen, Germany) with an HD-120 high-definition multi-leaf collimator (HD-MLC) was used for verifying the measured and calculated dose distributions for IMRT. The Eclipse 8.6 RTPS and anisotropic analytical algorithm (AAA) were used to calculate the radiation dose with a grid size of 2.5 mm and inhomogeneity correction. Our previous study [15] and Park et al. [16] showed that the AAA algorithm with a grid size of 2.5 mm is sufficiently accurate for low-density regions such as the lungs. Measurements were made by using I'mRT MatriXX with MULTICube, and the source-to-axis distance (SAD) was set to 100 cm. The calculated and measured radiation doses with I'mRT MatriXX were carried out at a 0g antry angle as stipulated by the internal QA protocol. Patient-specific head phantom and Gafchromic EBT2 film. The radiation dose was calculated in the same manner as that used with I'mRT MatriXX but on the sagittal plane because the radiation dose was to be measured in this plane. The calibration process for the Gafchromic EBT2 film was done in a previous study [17]. Among three channels (red, green, and blue), the red calibration curve was selected to compare the calculated and measured radiation doses for the patient-specific head phantom because the analog-to-digital (ADC) converter value for the red channel was least sensitive to changes in the radiation dosage. The calculation and measurement of the radiation dose with the 3D-printed patient-specific head phantom and Gafchromic EBT2 film were performed by using the fixed-gantry IMRT technique with 6 MV photons at gantry angles of 0˚, 40˚, 80˚, 120˚, 160˚, 200˚, 240˚, 280˚, and 320˚. In order to investigate the homogeneity of the HU of the head phantom, a point marked during the CT scan was designated as the center, as shown in Fig 4(B). Ten points were selected at the center, superior (+10 cm), and inferior (−10 cm) positions for HU measurement. S1 Table presents the results. The mean HU at the center was −340 with a standard deviation of 11.6, while the mean and standard deviation at the superior position were −338 and 14.6, respectively. The mean and standard deviation at the inferior position were −339 and 9.9, respectively. Overall, the HU for the patient-specific head phantom was largely homogeneous.  Fig 5(A) shows the calculated dose distribution at the frontal plane, while Fig 5(B) shows the measured dose distribution. Fig 5(C) is a gamma map in which the red regions indicate where γ ! 1, and the bluish-violet regions indicate where γ < 1. The acceptance DD and DTA used for gamma evaluation were set to 3% and 3 mm, respectively [4]. Fig 5(D) represents the gamma evaluation profiles. The red line represents the distribution of the measured dose, while the green line indicates the distribution of the calculated dose. The gamma passing rate (i.e., γ 1), was 97.28%, and the gamma failure rate (i.e., γ > 1) was 2.72%. Test for dose accuracy with the 3D-printed patient-specific head phantom    Fig 6(D) shows the gamma evaluation profiles using the 3Dprinted anthropomorphic patient-specific head phantom and Gafchromic EBT2 film in the sagittal plane. Similar to I'mRT MatriXX, the same values for the acceptance DD (3%) and DTA (3 mm) were applied in calculations with the patient-specific head phantom. The results produced a gamma passing rate (γ 1) of 95.97% and failure rate (γ > 1) of 4.03%.

Discussion
In this study, the feasibility of using a 3D-printed patient-specific head phantom to compare the calculated and measured radiation doses was verified. We fabricated an anthropomorphic patient-specific head phantom by using 3D printing technology to verify the calculated and measured radiation doses. We compared the conventional QA measured with a 2D diode array and the head phantom fabricated with a 3D printer.
While the radiation doses in radiotherapy treatments are often measured by using an ion chamber, diode 2D array, and films [4,5,7,8], there is little in the literature on utilizing a patient-specific phantom to calculate and measure radiation doses. In this research, a patientspecific head phantom was constructed from a DICOM file by using 3D printing technology, which has seen rapid advances in recent years.
In summary, the performance with the patient-specific head phantom was compared to that of I'mRT MatriXX, which is a well-established IMRT QA technique. The 2D array and patient-specific head phantom were analyzed individually by using the gamma index. At the acceptance DD and DTA values of 3% and 3 mm, respectively, the gamma passing rate with γ 1 was 97.28% with I'mRT MatriXX and 95.97% with the patient-specific head phantom for a difference of approximately 1.31%. Consequently, our results showed that the 3D-printed patient-specific head phantom has great potential for patient-specific QA of intensity-modulated radiotherapy.
In a similar study, Ehler et al. [8] created an anthropomorphic Rando phantom with heterogeneous tissue equivalence in order to compare the measured and calculated radiation doses, where FDM was applied to acrylonitrile butadiene styrene (ABS). They intended to realize dosimetric measurements that would not be possible in vivo and to avoid patient privacy issues during research. A nine-field IMRT plan was employed with a traditional verification method using the planar and cylindrical phantoms MapCHECK2 and ArcCHECK (Sun Nuclear Corp, Melbourne, FL). The mean percentage difference between the calculated and measured doses for the Rando phantom was 1.9%±2.8%, whereas the value with the 3D printed phantom was −0.1%±4.9%, so the difference between the two types of phantoms was minimal. Thus, they concluded that patient-specific phantoms fabricated by a 3D printer can be used for dosimetric verifications between the measured and calculated doses.
Joseph et al. [18] recently reported that polystyrene can be used to fabricate a 3D low-density phantom for radiotherapy QA with a relative electron density of 0.18-0.75 by using the ORION Delta 3D (SeeMeCNC Indiana) phantom. They concluded that 3D printing allows the fabrication of variable-density phantoms for radiotherapy QA in low-density regions from the correlation curve between the 3D printer infill value and relative electron density. However, they noted limitations for use in phantoms with an exceedingly low density.
These variable-density issues have been a great challenge in applying 3D printing technology to patient-specific QA of radiotherapy because most commercially available 3D printers currently support only one material at the time of fabrication. Thus, one limitation of our study was that we used only a single material (ABSplus) for the fabrication of the head 3D-printed patient-specific head phantom phantom, while the body of a real patient consists of various types of organs that produce a much wider range of HU values.
In future work, we plan to fabricate a variable-density patient-specific phantom to compare the calculated and measured doses. We also want to investigate the linear attenuation coefficient for radiotherapy by varying the different materials used in 3D printing.
Nevertheless, the 3D-printed phantom in this study can provide a better end-to-end test than the traditional quality control test.

Conclusions
In this research, a patient-specific head phantom identical to the head of a real patient was constructed with a 3D printer by using the patient's CT image. The phantom can be utilized for dosimetric verification that must be performed prior to complex radiotherapy treatment, such as IMRT and VMAT. In future work, we will research fabricating phantoms with variable density to simulate various components of the human body, including the lungs, fat, blood, muscle, soft tissue, and bones when comparing the calculated and measured radiation doses.