High Fidelity Anthropomorphic 3D Printed Models - Accuracy, Precision and Quality Control

The 3D modelling and printing of anthropomorphic 3D models based on CT are essential to modern anthropology and bioengineering. The accuracy and precision of those models are vital for their usability, which presents more challenges for their manufacturing and QA. In this study, three human humeri marked with X-ray-positive labels were CT scanned. The labels provided reference points for the subsequent measurements. The exported DICOM datasets were segmented in Slicer3D and 3D-printed with an FDM 3D printer with and without thermal deformation compensation. Three raters measured the bones and the 3D-printed models using a manual caliper. The DICOM measurements were conducted using the volume rendering module of Radiant DICOM viewer. The average distances between the reference points on the actual bones were used as ground truth to assess the accuracy of the DICOM dataset and the 3D-printed models. The average bias (AB) of the volume rendering measurements was
$-\mathbf{3}.\mathbf{24} \pm \mathbf{1}.\mathbf{13}\mathbf{mm}$
. The mean weighted bias (WB) was
$-\mathbf{24}.\mathbf{35}\pm \mathbf{9}.\mathbf{1}\boldsymbol{\mu} \mathbf{m}/$
mm. The mean weighted absolute difference (WAD) was 25.93
$\pm \mathbf{8}.\mathbf{85}\boldsymbol{\mu} \mathbf{m}/\mathbf{mm}$
. In the 3D-printed model without thermal compensation the AB was
$\mathbf{0}.\mathbf{11}\pm \mathbf{0}.\mathbf{27}\mathbf{mm}$
, the WB was
$-\mathbf{1}.\mathbf{89}\pm \mathbf{4}.\mathbf{68}\boldsymbol{\mu}\mathbf{m}/\mathbf{mm}$
, and the WAD was 7.85
$\pm \mathbf{4}.\mathbf{35}\boldsymbol{\mu}\mathbf{m}/\mathbf{mm}$
. The anthropomorphic 3D-printed models are an accurate instrument for the representation of the spatial properties of the skeleton. They can provide fast and reproducible results in morphometric studies on bone morphology. Thus 3D-printed models are a more reliable tool than the measurements of Volume renderings, which clinicians frequently use.*