Technical Note: Development of 3D‐printed breast phantoms for end‐to‐end testing of whole breast volumetric arc radiotherapy

Abstract End‐to‐end testing of a new breast radiotherapy technique preferably requires realistic phantom geometries, which is challenging to achieve using currently commercially available solutions. We have developed a series of three‐dimensional (3D)‐printed breast phantoms, with ionization chamber and radiochromic film inserts, which can be attached to a commercial anthropomorphic thorax phantom. A contoured left breast from a patient’s planning CT was mapped onto a CT of the CIRS E2E thorax phantom (CIRS Inc.) and cropped to fit the surface. Four versions of the breast were 3D printed, containing a cavity for an ionization chamber and slits for radiochromic film insertion in the three cardinal planes, respectively. The phantoms were fully compatible with surface scanning technology used for setup. The phantoms were validated using a whole‐breast volumetric modulated arc therapy protocol with a simultaneous integrated boost to the tumor bed (VMAT‐SIB). Six patient plans and one original plan on the breast phantom were verified with planar portal imaging, point dose, and film measurements in the MultiCube phantom and planar γ‐analysis using ArcCHECK diode array. Six patient plans were recalculated on the breast phantom (hybrid plans) and delivered with point dose and film measurements with 3% (local)/2 mm γ‐analysis. One complete end‐to‐end test on the breast phantom was performed. All plan quality verifications had point dose differences below 2.4% from the calculated dose and γ‐agreement scores (γAS) > 87.3% for film measurements in the MultiCube, portal dosimetry, and ArcCHECK. Point dose differences in the 3D‐printed phantoms were below 2.6% (median −1.4%, range −2.6%; 0.3%). Median γAS was 96.4% (range 80.1%–99.7%) for all film inserts. The proposed 3D‐printed attachable breast dosimetry phantoms have been shown to be a valuable tool for end‐to‐end testing of a new radiotherapy protocol. The workflow described in this report can aid users to create their own phantom‐specific breast 3D‐printed phantoms.


| INTRODUCTION
The validation of new volumetric arc radiotherapy (VMAT) or intensity-modulated radiotherapy (IMRT) protocols is generally performed via end-to-end tests, where the entire radiotherapy workflow from planning to delivery is verified. TG-119 recommends the use of target and structure geometries along with the target doses and dose constraints that are likely to be encountered in the clinic for IMRT commissioning. 1 However, most dosimetry phantoms lack the specific anatomy required for a "realistic" end-to-end test, namely, all steps from patient imaging, to contouring, to setup using surface scanning technology, for breast cancer plans.
Existing anthropomorphic phantoms with breast attachments are the Alderson Radiation Therapy phantom (Radiology Support Devices Inc., Carson, CA, US) and its earlier version, the Alderson RANDO phantom. However, they only contain cylindrical extrusions for thermoluminescent dosimeters (TLDs), not widely available in radiotherapy departments. Others, such as the end-to-end (E2E) SBRT Thorax phantom (CIRS Inc., Norfolk, VA, US), have multiple holes for ionization chamber (IC) and dosimetric film insertion, yet lack external features mimicking breasts. External features are also lacking in the thorax phantom of the Imaging and Radiation Oncology Core group (IROC) distributed for quality assurance (QA). 2 Moreover, all these phantoms have a shiny coating, cylindrical symmetry, or both, limiting the use of surface scanning technology used for patient setup, 3,4 which is increasingly being used for patient setup, 5 and whose positioning accuracy should be incorporated into commissioning.
Recent dispersion of low-cost three-dimensional (3D)-printing technology has allowed for the development of (patient) specific phantoms in radiotherapy. [6][7][8] This work describes the development of a series of end-to-end left breast attachments to the CIRS SBRT thorax phantom, using a mid-range 3D printer, allowing for commissioning of breast radiotherapy treatment techniques. All phantoms are fully compatible with surface scanning technology for accurate positioning and have either a film or IC insert centrally located in the breast. The phantoms were validated by performing hybrid plan measurementswhere patient plans are recalculated on the phantomfor six VMAT breast treatments with simultaneous integrated boost (VMAT-SIB). Additionally, an end-to-end test was performed whereby a VMAT-SIB plan was created on a simulation scan of the phantoms, the phantoms were positioned with surface scanning technology and the plan was delivered on the breast phantoms with film and IC insert.

2.A | Development of 3D-printed phantoms
To obtain a realistic breast shape, the contoured breast volume on the computed tomography (CT) of a patient was imported into 3DSlicer (version 4.10). 9 The breast volume was mapped onto a planning CT of the CIRS E2E SBRT thorax phantom, as shown in Fig. 1. The breast volume was then cropped to create a tight fit with the CIRS phantom. The phantom base shape of 698 cc was then exported as a stereolithography file (.stl). To create the final phantoms, 0.3-mm-wide slits were cut out using FreeCAD 10 (version 0.17) in the three cardinal planes (ie, sagittal, coronal, and axial). The fourth phantom was created by hollowing out a cylindrical cavity for the IC phantom (⌀ 6.75 mm).
The phantoms were printed in red polylactic acid (PLA) (ICE Filaments, Belgium) on a Raise3D N2 Plus 3D printer (Raise3D, The Netherlands) with 80% infill and two outlines. In a preparatory step cubes (5 × 5 × 5 cm 3 ) with varying infill were printed and scanned at 120 kVp, slice thickness 1 mm on a Somatom Sensation Open   | 317 exported at a resolution of 0.4 × 0.4 mm 2 . A global rescaling of the film to the TPS predicted dose and a low-dose exclusion threshold of 10% using a 3% (local)/2 mm γ-criterion, as per TG-218 recommendations, 18 were used. All plans were delivered on a Halcyon linac.

2.C | Phantom validation with hybrid plan deliveries
and an end-to-end test 2.C.1 | Hybrid plan verification on the E2E breast phantoms The six patient plans were recalculated on the planning CT of the E2E breasts, as a hybrid plan verification. Online setup was performed with AlignRT (VisionRT Ltd., UK) and verified with kV-CBCT, as per our image-guided RT (IGRT) protocol. All plans were verified using both A1SL point dose measurements and radiochromic film measurements. Radiochromic EBT3 films were cut with a Trotec laser cutter (Trotec Laser GmbH, Austria) using a template to match the inserts in the E2E breasts. All pieces were reassembled for scanning, as shown in Fig. S1 of the additional material.

2.C.2 | End-to-end test with a plan created on the E2E breast phantoms
An end-to-end test was performed by creating a seventh plan on the E2E breast. The planning CT with the IC E2E breast was delineated according to our departmental protocol, provided organs were present in the CIRS phantom. The following organs at risk (OAR) were delineated: contralateral breast/chest wall, heart, and lungs. The CTV 45.57Gy was the entire breast, cropped 5 mm below the skin. A boost volume CTV 55.86Gy was artificially drawn around the active volume of the IC, located centrally in the breast, as shown in Fig. 3.
PTVs were constructed by extending the CTVs by an isotropic 5 mm margin. A VMAT-SIB plan was generated using the strategy outlined in section B. Setup was again performed using AlignRT and verified by kV-CBCT, prior to treatment delivery. Ionization chamber and film measurements were performed.

3.B | Hybrid plan verification and end-to-end test on the E2E breast phantoms
Results of planar γ-analysis and point measurements in the E2E breasts are shown in Table 1. Median γAS was 96.4% (range 80.1%-99.7%) for all patients and film orientations, and median relative point dose difference was −1.4% from the predicted dose (range −2.6%; 0.3%).
The γ-analysis for the axial film of the end-to-end plan is shown in Fig. 4. The analysis for a hybrid plan verification on a coronal film is supplied in the additional materials.  The phantoms were fully compliant with the AlignRT surface scanning system, which is increasingly used for setup of breast cancer patients at radiotherapy departments. 5,19,20 The shape and red color of the PLA resulted in a stable surface registration result. During RTT training and hardware and software upgrade verification of the surface scanning system, we now use the E2E breasts compared to the vender-provided geometric phantom.
As a validation we chose VMAT-SIB plans due to the complex dose distribution with steep gradients, however, tangential field-infield or IMRT plans could also be used. We suspect a change in the planning protocol would have no effect on the acceptance of the phantoms in the clinic.
Previous 3D-printing dosimetry studies focused on developing a complete phantom, usually mimicking head and neck anatomy. 8,21,22 T A B L E 1 Dosimetric film and ionization chamber point measurements for the hybrid plans and end-to-end test. The authors were only able to find one other paper by Craft et al. 23 who created a phantom specifically for (postmastectomy) chest wall radiotherapy. In contrast, we aimed to only create an attachment to our in-house end-to-end phantom.

Case
Point doses differ from the calculated value to within 2.6% and film γ-analysis shows agreement scores above 80.1% for a VMAT-SIB protocol. We note generally lower agreement scores for the axial film inserts, compared to the sagittal and coronal insert. We suspect this effect to be due to a small air gap (1 mm) around the film due to the slightly larger insert width caused by local warping of the 3D print. In the future, the axial film phantom could be reprinted with care to eliminate warping. The low agreement score for the coronal film of the E2E breast, we suspect to be, due to a slight residual roll rotation of the phantom after setup, as the plan verification gamma agreement scores, using the MultiCube, EPID, and ArcCheck measurements, were above 88%.
We have limited the breast phantoms to a single size. It would, however, be useful to also create a supplementary set of phantoms for very large breasts (e.g., >1800 cc) or pendulous breast to verify the applicability of the VMAT-SIB technique for more extreme anatomies.
The proposed attachable left breast phantoms allow for a realistic (hybrid) end-to-end test of breast radiotherapy protocols. The procedure described in this report can be reproduced by others to create their own breast phantoms, matching in-house dosimetry equipment.

CONF LICT OF I NTEREST
This work was supported by Varian Medical Systems.

R E F E R E N C E S SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article. Figure S1. Assembled scanned film.