Methods

Design and 3D printing of “eco-friendly” target-specific passive scattering devices

The PS devices were designed to reduce the need for TS components by dividing them into 2 distinct parts. The reusable component consists of small metal and plastic spheres. In contrast, the TS hollow components were manufactured using standard 3D printing technologies, which helps to maintain relatively low costs. These hollow containers, either for collimators or compensators, were printed with polylactide (PLA, 25$/kg) filament with commercially available 3D printers, “Original Prusa i3 MK3S+” and “MK4S.”8 In addition, we performed one measurement printing the collimator bore with Copper-filled Metal Composite HTPLA filament,9 a PLA filament enriched with copper powder, commercially available and compatible with conventional filament printers, to test the impact of the material on the bore design.

The hollow container designs were created in Python for simple shapes and using 3D Slicer for the more complex case of a murine tumor. The designs were exported as Stereolithography binary format (STL) files and imported into Prusa Slicer.

The hollow containers were manually filled with polypropylene (PP) spheres10 to serve as compensator devices and with copper (Cu) spheres as collimating devices (PP: 1.191 mm radius, 0.007 g weight; Cu: 1.000 mm radius, 0.038 g weight). The spheres’ radii were selected to be small enough not to affect the dose distribution but not so small as to make handling difficult.

Experimental setup

All measurements were performed in PSI Gantry 2. In Table 1, we summarize the performed measurements and the different setups employed.

Measurement goal | Setup | Instrumentation | Measured quantity
Simulation versus measurement | (1) PLA hollow boxes with PP and Cu spheres. Single spot delivery with multiple energies (100-220 MeV). | QA phantom | DDC and trans-verse dose profile
Collimator performance | (2) PLA (and HTPLA) hollow collimators with Cu spheres, and a solid Cu collimator.Scattered 90-120 MeV beam. | CCD Camera | Transverse dose profile
Feasibility of PS delivery 1 | (3) PLA hollow collimator with Cu spheres and compensator with PP spheres designed for a spherical target.Scattered 120 MeV beam. | CCD Camera and PMMA slabs | Transverse dose profile
Feasibility of PS delivery 2 | (3) PLA hollow collimator with Cu spheres and compensator with PP spheres designed for a small tumor target. Scattered 110 MeV beam. | Gafchromic films in PMMA slabs | Longitudinal and transverse dose profile

To test the agreement between our simulation workflow and measurements, we performed beam size and depth dose curve (DDC) measurements using the PSI daily QA phantom. The phantom is equipped with 2 scoring planes to measure the dose profile perpendicular to the beam direction and a multilayer ionization chamber for the integral DDC. The multilayer ionization chamber consists of 128 plates with a water equivalent thickness of 2.3640 mm each. Gafchromic films interleaved in PMMA slabs. In Setup 2 and 3, we utilized a charge-coupled device camera attached to a scintillating screen to measure the beam profile after collimation (0.4 mm resolution). In Setup 3, we also measured the dose distributions using Gafchromic films interleaved in PMMA slabs.

Simulation workflow

All MC simulations in this work were performed using TOPAS (Tool for Particle Simulation,11, 12 Version 3.9), previously validated for PSI Gantry 2.13 We utilized the suggested default modular physics list11 (G4EMLOW6.48, G4NDL4.5, PhotonEvaporation3.2, RadioactiveDecay4.4, G4SAIDDATA1.1, G4NEUTRONXS1.4, G4PII1.3, G4ABLA3.0, G4ENSDFSTATE1.2.1, G4TENDL1.0) and the simulations were performed with 108 particles. The scorer had a resolution of 0.1 mm in the lateral direction.

The hollow containers were represented as simple PLA slabs or, for complex geometries such as the surface of the compensator for mouse irradiation, imported as STL files, as depicted in Figure 1 (Setup 3). The mouse CT was performed with a high-resolution small-animal CT camera used in a previous study,14 and the tumor contour was arbitrarily drawn in 3D Slicer.15 The target was converted into an STL file and exported. The compensator shape was defined by subtracting a simple STL box with the STL tumor volume using a Boolean difference operation, and then hollowing the resulting structure by removing its internal volume, leaving only the walls for printing the hollow container. To simulate the infill, we used and modified the function “pseudo gravity packing” from Porespy, Quantitative Image Analysis of Porous Materials.16 The original algorithm fills a 3D box with spheres by dropping each new sphere to the lowest point possible, simulating the effect of gravity. We modified this function to facilitate the filling of more complex shapes using the Open3D Python package17 and to enable the export of an appropriate file for simulations. TOPAS supports the inclusion of STL files, but even when undersampled, STL files representing thousands of spheres can be quite large. This can complicate the loading of geometry and the simulation of particle interactions, often resulting in lengthy loading times and extended execution durations for the simulations. Therefore, we have chosen to export the spheres directly as a list of TOPAS geometry components, specifically TsSphere.

Figure 1: TOPAS setup for the 4 measurements summarized in Table 1.