A 3D printed patient-specific vaginal mould for brachytherapy

Patient specific applicators are needed for vaginal brachytherapy treatments in cases where conventional cylindrical applicators are unsuitable or unusable. These applicators are often produced using a time-consuming and comparatively imprecise moulding method. This proof-of-concept study used Monte Carlo calculations to investigate potential dosimetric effects from creating applicators using several common 3D printing materials. A sample mould was then replicated using a fused deposition modelling (FDM) technique, which allowed catheter channels to be precisely placed with reference to treatment goals, before 3D printing from thermoplastic using a consumer-grade 3D printer. The Monte Carlo results indicated that several FDM filaments caused substantial dose depletions (up to 6%) within the model applicators while having a minor effect (less than 1%) on dose in surrounding tissue. Compared to the sample mould, the 3D printed applicator achieved superior dosimetry in terms of target coverage, while also passing manual tests of smoothness and usability. This study demonstrated an overall process by which 3D printing could replace an imprecise and time-consuming manual process and potentially achieve improved dosimetry in brachytherapy treatments of irregular vaginal anatomy.


Introduction
Brachytherapy is well-established as a treatment for gynaecological cancer [1,2], with intracavitary treatments for cervical cancer dating back to 1903 [1].High dose rate (HDR) brachytherapy treatments for vaginal cancers have conventionally used cylindrical applicators that are commercially available in varying diameters, with a single catheter through the centre of the cylinder.There are two major dosimetric limitations of this design.The first is that the single central catheter can produce only cylindrically symmetric dose distributions that produce sub-optimal treatments in cases where the required dose coverage is far from cylindrical [3].This limitation has been addressed by integrated shielding or the increasing use of cylindrical applicator designs that incorporate multiple catheters or the use of interstitial needles in combination with cylindrical applicators, so that comparatively complex dose distributions can be sculpted to cover specific regions [3,4].The second limitation is the cylindrical design of the applicator itself.Generic cylinders and rods commonly introduce air gaps and produce limited dose coverage of the target [5,6], especially for irregular anatomy such as for larger tumours or surgical resection sites [7,8].This substantial geometric limitation can require replacement of simple cylindrical applicators with entirely new and patient-specific designs [7,8].
As an alternative to commercial applicators with generic geometry, mouldable materials have been developed to produce a geometry specific to the patient anatomy [7,8].For example, Fricotan (Reid Technology, Auckland, New Zealand), a mouldable material designed for audiology applications, has been used to produce patient-specific moulded applicators that reduce the impact of air gaps while providing more optimised catheter placements [7].Optimisation of catheter placement is often achieved via an iterative fabrication-scanning-planning-refabrication process, where catheters are positioned in the mould by hand, with reference to the tumour shape, location and dimensions [7,8].Although the dosimetry is improved, the imprecise method of catheter placement within the mould will not guarantee a dose distribution that is fully optimised to the patient anatomy.Additionally, multiple attempts at optimising the source line placement can lead to resource wastage such as time and materials.
A further consideration when using moulded applicators is the potential non-water-equivalence of the moulding materials [7,9].This is a particular concern if the brachytherapy dose calculation formalism of the AAPM Task Group 43 is used, while uncertainties in material composition may also be a concern if model-based brachytherapy dose calculation algorithms are used [10,11].
These uncertainties in the radiological properties of the material and the imprecise placement of the source lines are trade-offs for reducing air gaps, increasing treatment accuracy and potentially increasing patient comfort [7] through the use of moulded patient specific applicators.This proof-ofconcept study therefore investigated a potential 3D printed alternative where catheter channels are precisely embedded in a material of known composition and density, to generate a more optimal brachytherapy plan, improving the plan quality further by increased coverage of the target volume.

Method
For this proof-of-concept study, Monte Carlo simulations were used to investigate the dosimetric effects of surrounding an Ir-192 brachytherapy source with several commonly used fused deposition modelling (FDM) 3D printing materials.EGSnrc and DOSRZ codes were used to calculate brachytherapy dose with and adjacent to cylinders of polylactic acid (PLA), polycarbonate (PC), polyethylene terephthalate glycol (PETG), acrylonitrile styrene-acrylate (ASA), and two Nylon filaments, Nylon and Nylon-X.All simulations used the nominal densities of the 3D printing materials, although simulations were repeated with reduced densities for PC and Nylon when samples of these two materials printed at 100% in-fill were found to have densities noticeably lower than nominal.Despite this issue, PC and Nylon were retained in this study due to their high melting points and likely ability to survive autoclave sterilisation.A 3.6 mm long line source was modelled in these simulations using the spectrum of an encapsulated Nucletron Ir-192 source, distributed with EGSnrc.Results were compared against calculations in an identical geometry, where the 3D printing material was replaced with water.
To demonstrate the proposed 3D printing workflow for gynaecological brachytherapy, a sample mould was replicated using an FDM 3D printing technique.The sample mould had been created using a fabrication process described by Nilsson et al [7], where a positive cast was created from Fricotan by an oncologist with gynaecologic brachytherapy experience, a negative was created using dental alginate, and then brachytherapy catheters were shaped into the desired configuration and supported by a small thermoplastic jig before being positioned in the negative and surrounded by Fricotan to produce the desired applicator design.Under this process, catheter positioning is constrained by the challenge of fixing the catheters in the desired shape without being able to visualise their position in the completed mould.
To create a 3D printed applicator with the same external geometry as the sample mould, the positive cast was CT scanned at high resolution (using a process previously described for 3D printed bolus quality assurance testing [12]) and imported into the MIM Maestro contouring software suite (MIM Software Inc, Cleveland, USA), where the Fricotan cast was contoured exported as a stereolithography (STL) structure.This structure as then smoothed using Meshmixer (Autodesk Inc, San Rafael, USA) and imported into Blender software (Blender Foundation, Amsterdam, Netherlands) where catheter positions were placed, based on the intended dose to the treatment target as specified by the clinician.Three catheters were placed within the structure, as this was the number used in the original mould.The resulting design was printed from PLA using an Ultimaker 2 Extended+ 3D printer (Ultimaker BV, Utrecht, Netherlands).Finally, the 3D printed applicator was CT scanned at highresolution so that a treatment plan could be created in the Varian Eclipse treatment planning system (Varian Medical Systems, Palo Alto, USA) and compared with the result from the sample mould.

Results
Figure 1 shows the results of the Monte Carlo calculations.The result for PLA is highlighted with filled black data points, since this material was selected for use in the workflow demonstration due to its low cost, broad availability and ease of use.The Monte Carlo result for PLA was similar to the results for ASA and PC (shown in Figure 1), with dose calculated within the applicator differing substantially from water although these differences had minimal effects on dose to surrounding tissue.Figure 2 shows key stages in the proposed process of fabricating and using a 3D printed applicator including: identification that a standard cylindrical applicator (Figure 2(a)) would be unsuitable for the patient, creation of a patient-specific mould (Figure 2(b)), use of the mould to design a 3D print with optimised catheter channel positions (Figure 2(c)), printing and use of the 3D printed applicator to achieve an improved treatment dose distribution compared to the manual method of physically placing catheters inside the mould (Figure 2(d)).In this case, the 3D printed applicator also passed a visual inspection and standard manual tests [7] for clinical suitability in terms of surface smoothness and easy insertion of catheters and clear transit of a dummy source.
The major limitation of this proof-of-concept was the use of PLA as the 3D printing material.PLA has a comparatively low melting point (often quoted as between 170 and 200 degrees), which makes it unsuitable for autoclave sterilization.Results in Figure 1 suggest that use of Nylon FDM filament might be advisable for producing improved water-equivalence while also being potentially sterilizable at high temperatures.However, the FDM printing process also has the potential to produce a very slight surface roughness which, despite feeling smooth to the touch, might represent an opportunity for increased bacterial growth.For these reasons, it is recommended that further work in this promising area focus on the use of high-resolution printing technologies such as stereolithography, PolyJet or selective layer sintering.Nonetheless, this study demonstrated an overall process by which 3D printing could replace an imprecise and time-consuming manual process and potentially achieve improved dosimetry in brachytherapy treatments of irregular vaginal anatomy.

Conclusion
The Monte Carlo results indicated that several FDM filaments caused substantial dose depletions (up to 6%) within the model applicators while having a minor effect (less than 1%) on dose in surrounding tissue.A dosimetrically superior plan was achieved using the proposed 3D printing process, highlighting the importance of catheter placement in brachytherapy.In addition to providing the patient with a treatment plan optimised to their anatomy, a 3D printing process combined with the educated, precise placement of catheters can conserve time and materials, potentially streamlining the patient journey.

Figure 1 .
Figure1.Dose calculated within and around a 3D printed applicator, relative to the dose calculated in water, using a worst-case thickness of 2 cm of 3D printed material between the Ir-192 source and the surrounding tissue, modelled as water.

Figure 2 .
Figure 2. (a) Photograph of cylindrical applicator.(b) Photograph of moulded patient-specific applicator.(c) 3D print design with internal catheter geometry.(d) Dosimetric comparison between Fricotan mould (left) and 3D printed applicator (right), where the cyan structure represents the treatment volume to be irradiated by the prescribed dose (blue) and the 90% coverage dose (magenta).