3D printing of the radiation modifier

During target segmentation, a bolus structure was created and a density override was applied to the bolus structure in the dose calculation. The bolus structure was designed to allow a 1 mm margin from the skin to bolus to compensate for fur that is not apparent on the CT scan. The bolus structure was exported from the segmentation software as a DICOM-RT file and converted to a .stl (stereolithography) file. The .stl file was loaded in 3D modeling software (Netfabb, Autodesk Inc, San Rafael CA). The 3D model was inspected for proper conversion from DICOM-RT format to .stl and if necessary, modification to the model was performed. All models were smoothed for 5 iterations with “prevent model shrinking” set to off.

For all cases, the plastic used was Polyethylene Terephthalate–Glycol-Modified, PETG, (eSun, Shenzhen Esun Industrial Co. Ltd., Shenzhen, China). Printing was performed on a Fused Deposition Modelling type 3D printer (Taz 6, Aleph Objects, Inc., Loveland CO). Initially, a nozzle with a 0.5 mm extrusion diameter was used (Case 1 through Case 11) and later, for Case 12 through Case 14, the nozzle was upgraded to 1.2 mm extrusion diameter (MOARstruder, Aleph Objects, Inc., Loveland CO). The larger extrusion diameter allowed for reduced print times due to two factors: faster extrusion through the larger diameter opening and larger layer heights resulting in fewer printing layers. The layer heights were 0.3 mm for the 0.5 mm nozzle and 0.6 mm for the 1.2 mm nozzle. The bolus for Case 11 was originally printed with the standard nozzle and after the upgrade the bolus was reprinted with the upgraded nozzle in order to compare print times for the same 3D model. All printing parameters were kept constant when possible for the 0.5 mm and 1.2 mm nozzle print settings. To that end the standard print settings supplied with the printer for nGen (a proprietary PET-G) were used with minor modifications for both extruders.

The time required to generate each 3D printed bolus was recorded for all cases. The process was divided into the following subroutines: bolus segmentation, conversion from DICOM-RT structure to 3D model (including model processing and smoothing), and 3D print time. Time reporting was limited to 0.25 hour (h) increments except for 3D print time, which was recorded by the printer precisely to the minute.