Commercial filament testing for use in 3D printed phantoms

There is a great demand for phantoms by many areas of knowledge to be used for teaching or daily work. However, commercial phantoms are expensive and hard to obtain, especially in countries going through development. As an alternative, 3D printing can be the way to produce less expensive and reliable 3D phantoms. The goal of this study is to evaluate 14 available commercial filaments, in order to find if and how they can be used in 3D printed phantoms in computed tomography. Each material was printed as a 2 cm edge cube with rectilinear pattern and 60, 80 and 100% infill. The 80% infill of five other patterns were also printed and compared. Each 100% infill cube was weighted and had its density calculated. After that, the cubes were scanned in a Philips CT Brilliance 6 with 120 kVp, 200 mA, 2 mm slices and standard reconstruction. At the center of each cube, a similar to 120 mm(2) region of interest was set to measure the mean Hounsfield Unit (HU) and its standard deviation. The software Origin was used to plot HU results for rectilinear pattern, determine linear trends with its R-2 and compare achieved values with HU tissue range from literature. To confirm the response of HU values of selected tested materials in CT imaging as a function of percentage infill, a phantom prototype of a finger was 3D printed. The HU of the tested materials ranged from -516.2 +/- 7.3 to 329.8 +/- 18.9. All human tissues could be mimicked making use of these materials, except cortical bone above similar to 350 HU and tooth parts. The most promising filament was PLA + Cu, due to the multiple infill configuration that allows the resulting HU range to represent from adipose and skin tissue to marrow bone.