Investigation of the effects of spinal surgical implants on radiotherapy dosimetry: A study of 3D printed phantoms.

Abstract

The use of three-dimensional (3D) printing to develop custom phantoms for dosimetric studies in radiotherapy is increasing. The process allows production of phantoms designed to evaluated specific geometries, patients, or patient groups with a defining feature. The ability to print bone-equivalent phantoms has, however, proved challenging. The purpose of this work was to 3D print a series of three similar spine phantoms containing no surgical implants, implants made of titanium, and implants made of carbon fiber, for future dosimetric and imaging studies. Phantoms were evaluated for (a) tissue and bone equivalence, (b) geometric accuracy compared to design, and (c) similarity to one another. Sample blocks of PLA, HIPS, and StoneFil PLA-concrete with different infill densities were printed to evaluate tissue and bone equivalence. The samples were used to develop CT to physical (PD) and effective relative electron density (REDeff ) conversion curves and define the settings for printing the phantoms. CT scans of the printed phantoms were obtained to assess the geometry and densities achieved. Mean distance to agreement (MDA) and DICE coefficient (DSC) values were calculated between contours defining the different materials, obtained from design and like phantom modules. HU values were used to determine PD and REDeff and subsequently evaluate tissue and bone equivalence. Sample objects showed linear relationships between HU and both PD and REDeff for both PLA and StoneFil. The PD and REDeff of the objects calculated using clinical CT conversion curves were not accurate and custom conversion curves were required. PLA printed with 90% infill density was found to have a PD of 1.11±0.03g.cm-3 and REDeff of 1.04±0.02 and selected for tissue- equivalent phantom elements. StoneFil printed with 100% infill density showed a PD of 1.35±0.03g.cm-3 and REDeff of 1.24±0.04 and was selected for bone-equivalent elements. Upon evaluation of the final phantoms, the PLA elements displayed PD in the range of 1.10±0.03g.cm-3 -1.13±0.03g.cm-3 and REDeff in the range of 1.02±0.03-1.06±0.03. The StoneFil elements showed PD in the range of 1.43±0.04g.cm-3 -1.46±0.04g.cm-3 and REDeff in the range of 1.31±0.04-1.33±0.04. The PLA phantom elements were shown to have MDA of ≤1.00mm and DSC of ≥0.95 compared to design, and ≤0.48mm and ≥0.91 compared like modules. The StoneFil elements displayed MDA values of ≤0.44mm and DSC of ≥0.98 compared to design and ≤0.43mm and ≥0.92 compared like modules. Phantoms which were radiologically equivalent to tissue and bone were produced with a high level of similarity to design and even higher level of similarity of one another. When used in conjunction with the derived CT to PD or REDeff conversion curves they are suitable for evaluating the effects of spinal surgical implants of varying material of construction.