Results

Simulations versus measurements

As a first step of the analysis, we evaluated whether we could correctly reproduce the “eco-friendly” devices with TOPAS MC simulations. With the modified function “pseudo gravity-packing,” we generated one 1.3 cm height box filled with Cu spheres and one 2.3 cm height box filled with PP spheres. We printed a PLA box of the given size, manually filled it with the spheres, and measured the integrated DDC and the beam profile for multiple energies between 100 and 220 MeV (Setup 1). We compared the TOPAS simulation with the measurements for 2 example energies in Figure 2. The range (R80) and beam lateral profile for Cu are within a few percent (relative difference 0.5% and 3%, respectively) and fall within the detector resolution (2.364 mm in depth and 2 mm in the transverse profile). Compared to the pristine Bragg peak, the range is reduced by 4.7 cm (4.8 cm in simulations), and the beam size increases by 2.5% (6% in simulations). For PP, the difference in range is within 5%, with a range decrease of 1.7 cm (1.3 in simulations) compared to the non-degraded DDC. The measured lateral profile aligns with the simulated one, with a relative difference of 10%. The discrepancy between measured and simulated values may be attributed to differences in how the beam lateral profile is modeled in TOPAS, or to an inaccurate specification of the material properties of PLA.

Figure 2: (Left) Experiment setup to measure with the PSI daily QA phantom the DDC and lateral dose profile of the beam passing through simple hollow boxes filled with PP or Cu spheres. (Right) Integral DDC and beam size: comparison between TOPAS simulation (full line) and measurements for a box filled with Cu (top) or PP spheres (bottom).

Collimator performance

We measured and compared the dose profile of a monoenergetic scattered beam after collimating it with an “eco-friendly” collimator and a solid copper collimator (Setup 2). The solid collimator was a copper device of 4 x 4 x 1.5 cm, with an aperture of 1.6 cm in diameter, and the printed PLA hole container was 4 x 4 x 1.7 cm to account for the air space between the spheres. The printing time was less than 20 minutes, and 4 g of PLA filaments were needed.

Figure 3(A) shows a comparison between measurements and simulations for a 90 MeV scattered beam. The difference between the measured and simulated width is within a few percent (1%-3%) for both scenarios. At high energies, we observe that the PLA bore affects the dose distribution, producing a dose halo, see Figure 3(B, PLA bore). This is due to particles not being stopped by the collimator as they pass through the PLA-defined bore. To account for that, we evaluated 2 options. We simulated the PS setup using a collimator in which the bore is defined by copper instead of plastic and by doubling the height of the collimator. In both cases, we reduced the halo dose. However, while the collimator in which the bore is defined by copper effectively prevents all particles from passing through, some particles are still able to pass through the PLA-defined bore despite it being longer. Further, we experimentally tested the superior performance of the copper-bore collimator by 3D-printing the collimator with an HTPLA insert defining the bore, a PLA filament enriched with copper powder, as no full copper filament was available. Results are shown in Figure 3(B, HTPLA bore).

Figure 3: (A) Dose profile: comparison between TOPAS simulation (dashed line) and measurements for scattered beam collimated with an “eco-friendly” collimator (blue) and a solid copper collimator (red). (B) CCD measured 2D dose distribution for an “eco-friendly” collimator with an HTPLA (top) and an PLA (bottom) bore; (C) Simulated dose profile with a PLA bore (red), copper bore (blue), and a longer collimator (green) compared to the corresponding full solid collimator (dotted black).

Target-specific PS delivery

To test a more experimentally relevant scenario, we designed a collimator and a compensator to irradiate a 2 cm-diameter spherical target. We filled the collimator with Cu spheres and the compensator with PP spheres. The required filament was less than 10 g, and the total printing time was 30 minutes. We simulated and measured (Setup 3) the dose profile for 120 MeV at 2 different depths, as shown in Figure 4(A), and found a mean squared error of less than 1%.

Figure 4: “Eco-friendly” collimator and compensator for (A) a spherical target: (Left) 3D printed collimator and compensator with spheres, (Center) simulated dose profile, (Right) comparison between simulated and measured dose profile (B) a murine brain-like tumor: (Left) 3D printed collimator and compensator with spheres, (Center) mouse CT with contour of the target (volume = 157 mm3); (Right) Longitudinal and transversal dose measurements with dose profile.

To qualitatively evaluate the feasibility of our method for more complex shapes, we designed an experimental setup to irradiate a murine brain-like tumor. We designed the collimator and compensator to irradiate the targeted area and 3D-printed them. The measured (Setup 3) dose distributions are shown in Figure 4(B).