A framework for clinical commissioning of 3D‐printed patient support or immobilization devices in photon radiotherapy

Abstract Purpose The objective of this work is to outline a framework for dosimetric characterization that will comprehensively detail the clinical commissioning steps for 3D‐printed materials applied as patient support or immobilization devices in photon radiotherapy. The complex nature of 3D‐printed materials with application to patient‐specific configurations requires careful consideration. The framework presented is generalizable to any 3D‐printed object where the infill and shell combinations are unknown. Methods A representative cylinder and wedge were used as test objects to characterize devices that may be printed of unknown, patient‐specific dimensions. A case study of a 3D‐printed CSI immobilization board was presented as an example of an object of known, but adaptable dimensions and proprietary material composition. A series of measurements were performed to characterize the material's kV radiologic properties, MV attenuation measurements and calculations, energy spectrum water equivalency, and surface dose measurements. These measurements complement the recommendations of the AAPM's TG176 to characterize the additional complexity of 3D‐printed materials for use in a clinical radiotherapy environment. Results The dosimetric characterization of 3D‐printed test objects and a case study device informed the development of a step‐by‐step template that can easily be followed by clinicians to accurately and safely utilize 3D‐printed materials as patient‐specific support or immobilization devices. Conclusions A series of steps is outlined to provide a formulaic approach to clinically commission 3D‐printed materials that may possess varying material composition, infill patterns, and patient‐specific dimensions.


| INTRODUCTION
Three-dimensional (3D) printing technology has substantially improved over the past decade and is already being used within radiotherapy for a variety of applications, including bolus, 1-3 brachytherapy applicators, 4 quality control and anthropomorphic phantoms, [5][6][7][8] and preclinical immobilization. 2,9 Technology variations across 3D printer vendors introduce different material compositions and print infill patterns that are often proprietary and possess unknown radiologic properties for use in radiotherapy. This contrasts with most commercially manufactured immobilization devices specifically designed for radiotherapy where it is in a vendor's best interest to provide radiotherapy departments with detailed device specifications.
The AAPM's TG176 reported on dosimetric effects caused by couch tops and immobilization devices. 10 The manuscript included a comprehensive list of existing literature that characterizes the dosimetric properties of common, commercial radiotherapy devices. With a 3D printer, objects can be individually customized by the clinic to accommodate a specific patient's treatment and, therefore, it is not feasible to comprehensively characterize every device, material, and geometry configuration that may be manufactured for patient use. This is especially true with 3D printing applications designed for patient specific anatomy, which may require printed devices that possess patient-specific configurable arrangements or are disposable.
For example, characterization of a 3D-printed device's attenuation at varying beam obliquity is not practical when a device's thickness is not known prior to modifying the design for patient specific anatomy.
Additionally, TG176 primarily characterizes patient support and immobilization devices through attenuation and surface dose effects.
TG176 serves as the foundation for understanding the dosimetric impact of traditional radiotherapy devices, but additional information is required for 3D-printed materials that is dependent on proprietary material composition and print infill pattern.
In this manuscript, a framework is provided for dosimetric characterization of 3D-printed materials that may be utilized without fixed device dimensions. This work builds upon the recommendations of AAPM's TG176 to include additional characterization techniques specifically for 3D-printed materials. An immobilization device to be used clinically in volumetric-modulated arc therapy (VMAT) cranio-spinal irradiation (CSI) is used as a case study to demonstrate the results of the outlined methodology. The characterization experiments that align with AAPM's TG176 are reported in this work in the context of performing the measurements on the sample 3Dprinted device. As well, the additional characterization measurements outlined specifically for 3D-printed materials are described in detail.

| METHODS
All measurements were performed with a proprietary material (Onyx ™ ) that was printed from the Markforged Onyx One (Markforged, Boston, MA, USA). Onyx ™ is a carbon fiber-based material with a nylon additive, of proprietary composition. The 3D-printed objects used in the following experiments were printed with a nominal 50% infill and 1-mm shell thickness. The framework presented is applicable to any 3D-printed object where the infill and shell combinations are unknown. All linear accelerator-based measurements were performed on Varian accelerators (Varian Medical Systems Inc., Palo Alto, CA, USA).

2.A | Representative 3D-printed material
To characterize the 3D-printed material for general use in radiotherapy, two representative test objects were used: a cylinder [ Fig. 1

(a)]
was printed with 15-cm diameter and 3-cm height and a wedge (not shown) with 3-cm height and 10-cm length. The representative cylinder diameter was selected so that a reference 10 × 10 cm 2 field would fit entirely over the diameter of the device. The cylinder height was selected so that the ratio of infill pattern to shell thickness is clinically similar to the intended use of the clinical immobilization device. The representative wedge height was selected to match the height of the cylinder. The density of the representative cylinder was estimated to be 0.456 g/cm 3 , which is the quantity used to estimate the water equivalent thickness for experimental analysis.

2.B | Case Study: VMAT CSI immobilization board
This CSI immobilization board [ Fig. 1(b)] has been designed so that it is highly adaptable by easily movable sets of interlocking wedges along a thin 3D-printed grooved surface. The movability of the wedges allows the same immobilization device to be used for the

2.C | Determination of kV properties
The representative cylinder was used to assess kilovoltage (kV) radiologic properties of Onyx ™ . The cylinder was imaged at both a high  Table 1.

2.D | Attenuation comparison to treatment planning system
To characterize attenuation of the Onyx ™ material for the representative cylinder and CSI immobilization board, three experimental setups were used. First, 10 cm of SolidWater ® (Best Medical, Springfield, VA, USA) with a Capintec Farmer ionization chamber (Radiation Products Design, Inc, Albertville, MN, USA) placed at 5-cm depth within the SolidWater ® was aligned to the beam's central axis.
The surface of the SolidWater ® was aligned to 95-cm source-surface distance (SSD). A 10 × 10 cm 2 field was used to deliver 100 monitor units (MUs) for all three beam energies that have been clinically commissioned for VMAT CSI: 6 MV, 6 flattening filter free (FFF), and 10 FFF. The same measurement was repeated for all beam energies with the representative cylinder placed atop the SolidWater ® and again with the cylinder removed and replaced by the CSI immobilization board. A CT image of the experimental setup described with cylinder atop SolidWater ® is shown in Fig. 2.
The ratio of charge collected with and without the cylinder was computed to report attenuation per centimeter of Onyx ™ . The ratio of charge collected with and without the CSI immobilization device was also reported to determine the CSI immobilization board attenuation in a representative treatment arrangement.
A CT image of the three experimental setups used to measure Onyx ™ attenuation was acquired. The CT images were imported to Eclipse version 13.6 (Varian Medical Systems, Palo Alto, CA, USA) where the experimental setup could be recreated within the treatment planning system. An anterior field was created with the field isocenter and treatment plan's prescription point set to 5-cm depth (i.e., the ionization chamber location). A prescription dose of 1000 cGy was assigned to the prescription point and dose was calculated with Analytical Anisotropic Algorithm (version 13.6.23, Varian Eclipse). The ratio of monitor units (MUs) with and without the representative cylinder as well as with and without CSI immobilization board was computed.
The ratio of the MUs required to deliver the prescription dose is compared to the ratio of the charge collected by the experimental ionization chamber experiment with the same setup. These results are used to assess the clinical treatment planning system's dose calculation accuracy when the 3D-printed material is in the beam's path.

2.E | Energy spectrum water equivalency
Reference percent depth dose (PDD) curves were acquired using a 3D Scanner ™ (Sun Nuclear, Melbourne, FL, USA) automatic scanning T A B L E 1 CT imaging parameters for high and clinical-resolution images.

CT Parameter
High resolution  The films were calibrated using the methodology outlined by Morrison et al. 11 To characterize the relative change in surface dose between the two scenarios, films profiles were analyzed in Matlab (Mathworks, Natick, MA, USA).

3.A | Determination of kV properties
The ROI measurements on the CT image of the representative cylinder yield a mean (standard deviation) HU of −637 (11.7) and −606 F I G . 3. Surface dose measurements using GAFchromic ™ film. (a) film directly on couch, and b) film on CSI immobilization board, and (c) film beneath the representative wedge, and (d) film beneath the representative cylinder. In all scenarios, the film is placed at isocenter and a 10 × 10 cm 2 field is delivered with 300 MUs. In (a) and (b), the gantry angle is 180 degrees while in (c) and (d), the gantry angle is 0°.

3.B | Attenuation comparison to treatment planning system
The measured and calculated dose for the representative cylinder and CSI immobilization board are reported in Table 2. The attenuation per centimeter is reported for the representative cylinder as it provides a measure of the attenuation that could be expected for a device with variable thickness. The total attenuation for a single point measurement is reported for the CSI immobilization board.

3.C | Energy spectrum water equivalency
The measured PDDs for the 6-MV beam are shown in Fig. 4. The raw data for the reference PDD and Onyx-based PDD are shown in Fig. 4(a). In Fig. 4(b) the shifted (1.368 cm) and ISL corrected Onyxbased PDD is shown alongside the raw reference PDD.
The same measurements and analysis were performed on 6 and 10 FFF beams (not shown). The corrected Onyx-based PDD shows good agreement with the reference PDD, indicating that a water equivalent estimate of the Onyx ™ material is reasonable.

3.D | Surface dose measurements
A plot of the absolute dose measurements comparing the representative wedge to the representative cylinder are shown in Fig. 5(a).
The wedge profile at maximum thickness (3 cm The absolute dose of the film measurements on the couch surface and CSI immobilization board surface are plotted in Fig. 5(b).
The mean values of the dose profile for the couch dose and CSI dose are 251.2 and 292.5 cGy, respectively, which is a relative increase in dose of approximately 16%.

| DISCUSSION
The customizability and simplicity of 3D printing aligns with radiotherapy aims of creating patient-specific immobilization devices.
With the recent increase in radiotherapy 3D-printing literature, a likely trend towards greater integration of 3D printing in radiotherapy applications requires a comprehensive framework for clinical implementation of the devices. The presented framework builds on the existing scope of TG176, 10 providing the additional measurements and considerations that are required for 3D-printed immobilization devices compared to commercially manufactured radiotherapy devices. A summary of the proposed commissioning procedure, as illustrated in this manuscript, is as follows.
1. Estimate the water equivalent thickness of the print material.
2. Print two representative devices to be used as test objects: T A B L E 2 Relative dosimetric difference in Onyx ™ attenuation. A comparison between measurement and calculation for the representative cylinder and CSI immobilization board.

Beam energy
Representative cylinder (attenuation percent per cm) CSI immobilization board (total attenuation percent)  3. Perform attenuation measurements with the cylinder.

4.
Perform surface dose measurements with the cylinder and wedge.

5.
Acquire CT images of the attenuation experiment geometry and import into the treatment planning system to assess dose calculation accuracy. Compare the attenuation measurements in step (3) to the attenuation calculations in the treatment planning system. A unique aspect of 3D printing is that objects are typically manufactured with a thin shell as well as an infill density and pattern that is user defined (honeycomb, pillars, etc.). 7,12 Although the measurement of density may appear to be straightforward, the relationship between shell thickness and infill density as they contribute to physical density vary depending on the printed device.
The precision of the density in a printed device and its effect on dosimetry requires investigation before clinical implementation. The purpose of the wedge measurements is to ensure the water equivalent thickness is valid for the range of densities and thicknesses to be used.
The dosimetric behaviour of the infill pattern was characterized for specific materials in application as bolus by Ricotti et al. 3 and found that for infill percentages greater than 20%, the material could be approximated as homogeneous. However, this relationship is not known for other 3D-printed materials, which highlights the importance of surface dose measurements that may yield a dosimetric pattern that is reflective of the print pattern. In Fig. 5(a) appropriately estimated with a water equivalent thickness. Though PDDs are relatively insensitive to a change in energy spectrum, agreement in the reference and material-based PDDs provide the clinician confidence that the dose calculation algorithm will accurately compute dose in the material. Methodologies for estimating electron density from CT number have been previously explored. 13,14 Notably, Michiels et al. 2 reported on the use of dualenergy CT to estimate the effective atomic number of a 3D-printed material. However, this is not practical for routine clinical implementation but could be used as an investigational tool. The surface dose and PDD measurements are designed to provide further evidence that the material properties are well understood in the radiotherapy context.
This work outlines the characterization framework required for use of 3D-printed materials as an immobilization device. For use as bolus, these dosimetric measurements are useful but the additional requirement to assess the intended clinical device's reproducibility in patient setup should be evaluated. Often, the rigidity of the 3D-printed materials may limit the application to malleable bolus devices. Furthermore, the case study results presented in this work do not assess the consistency in print, which could be important if devices are intended for disposable use. In that scenario, multiple test objects and a relevant subset of these recommended dosimetric experiments could be repeated to understand the variation observed across devices.

| CONCLUSION
A framework is presented for dosimetric characterization of 3D-printed patient support and immobilization devices. The study performs measurements on representative test objects and uses a case study of a 3D-printed CSI immobilization board to outline requirements for clinical commissioning of the 3D-printed material.
Recommendations are made to clearly define a series of steps that can be implemented by a user with access to generic 3D printers and materials.