2.1 PLA characterization
3D printed phantoms were created using Polylactic acid (PLA) (PLA Vanilla white, PRM-PLA-WHT-1000, Prusa Research, Prague Czech Republic). PLA has a backbone formula C3H402 and is derived from plant starch such as corn, with a mass density of 1.24 g cm − 1. The average density of the printed object can be adjusted by adjusting the infill. Adjusting the infill essentially creates submillimetre air gaps in the print that are not detectable on the CT scanner. Various 4 × 4 × 4 cm3 cubes of different infill from 90–100% were printed and CT scanned using a Toshiba Aquilion LB CT scanner (Canon medical systems, Otawara, Tochigi, Japan) to create a density calibration curve. All phantoms were then printed with an infill to achieve a mass density closest to 1 g cm − 3.
To verify the water equivalence of this material, 15 cm × 15 cm solid slabs of 0.1, 0.2, 0.5 and 1 cm thickness were created (PLA Slabs). The dose for a 6 cm × 6 cm electron applicator at SSD 100 cm and at depths of 1.4 cm, 2.1 cm and 2.9 cm for 6 MeV, 9 MeV and 12 MeV beams respectively were measured with an NACP chamber (SN 20081, IBA Dosimetry, Germany) and compared to solid water measurements at the same depth.
2.2 3D printed curvature phantoms
Both spherical and cylindrical curvature phantoms were created (Fig. 1). The spherical phantoms mimic treatment sites such as the scalp, breast and nose, and cylindrical the limbs and torso. The curvature phantom models were created using Fusion 360 (Autodesk, Inc. CA, USA), sliced in PrusaSlicer (version 2.3.3, Prusa Research, Prague, Czech Republic) and 3D printed from PLA using the Prusa i3 MK3S 3D Printer (Prusa Research, Prague Czech Republic). Print times raged from 20 min to 4 hours. Each phantom was sliced into 3 parts so that each additional part corresponds to dmax of each electron energy available in the department (Fig. 2). These were 1.4 cm, 2.1 cm and 2.8 cm for the 6 MeV, 9 MeV and 12 MeV electron energies, respectively. The spherical phantoms had radii of curvature, \({r}_{c}\), of 2.9. 3.9, 4.9, 6.3, 7.5 and 10cm. Additional phantoms with \({r}_{c}\) of 1.4 cm for the 6 MeV electrons, and \({r}_{c}\) of 2.1 cm for the 6 and 9 MeV electrons, were also created. The cylindrical phantoms had \({r}_{c}\) of 2.9, 3.9, 4.9 and 7.5 cm.
2.3 Contour factor measurements
Measurements were made on 3 beam matched Varian Truebeam Linear accelerators (Linac) (Varian medical systems, Palo Alto, CA, USA).
An NACP chamber (SN 20081, IBA Dosimetry, Germany) was placed in solid water (SP34, IBA Dosimetry, Germany). For the reference reading, solid water of equal thickness to the curvature phantom was added on top of the chamber. For the phantom readings, the solid water was removed, and the curvature phantom placed directly on top of the chamber, centred in the field light crosshairs (Fig. 3).
A 10 × 10 cm2 electron applicator and a source to surface distance (SSD) of 100cm was used for all the measurements. The measured CCF, \(CC{F}_{meas}\) is then the ratio of the electrometer reading with curvature phantom to the reference reading in solid water.
Measurements were also made with the 6 × 6 cm2 and 20 × 20 cm2 applicators on a single Linac to investigate the effect the field size has on \(CC{F}_{meas}\).
2.4 Calculation of CCFs in the TPS
Virtual spherical and cylindrical phantoms of the same radii as the phantoms were created in eclipse (v 15.6, Varian Medical Systems, Palo Alto, CA, USA). A solid cube phantom was created so that the surface is line with the depths as indicated above, with the curved contour placed on top. The density was set to that of water (-2 HU in Eclipse). A 10 cm × 10 cm applicator was used with the SSD set to 100 cm and monitor units set to 100 MU. EMC calculations were computed with a statistical uncertainty of 1%.
An Eclipse Scripting Application Program Interface (ESAPI v15.6) script was written to take the average dose in a 1 cm diameter plane located 1 mm deeper than dmax depth (1.5 cm, 2.2 cm and 3 cm), corresponding to the NACP chamber position (Fig. 4). This is the dimensions of the NACP chamber, with effective point of measurement of 0.6 mm (inner surface of front window). The ratio of this to an EMC calculation made on a flat phantom is then the CCF calculated in the TPS, \(CC{F}_{TPS}\).
2.5 Application of CCFs to patient cases
An ESAPI script was written to interpolate the patient CCF from \(CC{F}_{Meas}\) values based on the beam geometry and the patient external body contour. To do this, the script samples the patient body contour along the inline and crossline beam axis within the extent of the applicator size. A circle is then fitted to each curve with the radius of the circle fit equal to \({r}_{c,y}\) and \({r}_{c,x}\) in the inline and crossline dimensions respectively. If the centre of the circle is outside the patient body contour than the patient is assumed flat as concave surfaces were not measured in this study. \({r}_{c}\) can also be estimated by eye using the circular cursor tool in eclipse, by adjusting the radius of the tool and matching it to the patient CT image. Both were used to verify each other.
For clinical use, an assumption was made that \(CC{F}_{meas}=1\) at \({r}_{c}=30cm\). I.e., any patient contour with \({r}_{c}\ge 30cm\) is assumed flat. If both \({r}_{c,x}\le 30cm\) and \({r}_{c,y}\le 30cm\), then the average \({r}_{c}\) is used for interpolation of \(CC{F}_{Meas, Sph}\). If only one of \({r}_{c,x}\)or \({r}_{c,y}\) is greater than 30cm, then the smaller of \({r}_{c,x}\) or \({r}_{c, y}\) is used to interpolate from \(CC{F}_{Meas, Cyl}\).
This factor was applied to past patient IMU calculations to correct for patient curvature. 14 previous electron patients that had IMU calculations differing from the TPS MU by greater than the departments 5% tolerance were recalculated using CCFs. The CCF was calculated using both the ESAPI script and by assessing the patient using the circular cursor tool to estimate curvature by eye. The IMU calculation was made in ClearCalc IMU software (Version 2.0.13, Radformation, Inc. New York, NY, USA), and the CCF entered as an optional factor.
2.6 Clinical implementation
These contour factors were implemented into the department. Before implementation, the plots were smoothed, and outlying data points removed. An ESAPI script is used for automatic CCF calculation. This calculated factor is then entered into the optional factor section of ClearCalc.