An investigation into 3D printing of osteological remains: the metrology and ethics of virtual anthropology

3D printed replicas of human remains are useful tools in courtroom demonstrations. Presently, little published research has investigated the surface quality of printed replicas for use in the presentation of forensic anthropology evidence. In this study, 3D printed replicas of nine human bones were reconstructed from computed tomography (CT) scan data using selective laser sintering (SLS). A three-phased approach assessed: i) the metric accuracy of the 3D prints; ii) the viability of applying age and sex estimation methods (with multiple observers (n = 8); and, iii) the surface quality using a customized scoring method (with multiple observers (n = 8)). The results confirmed that the prints in this study were accurate to within 2.0 mm of the original dry bone. Observers were able to confidently assess the gross features of the prints; however fine surface details were not always well represented compared to the dry bones. These findings confirm the applicability of 3D printed replicas for courtroom exhibition of gross features but offer caution against their use when fine detailing is important for evaluative interpretation.

There is currently no published empirical evidence‐base demonstrating 3D printing to be an accurate and reliable tool in forensic anthropology, despite 3D printed replicas being exhibited as demonstrative evidence in court. In this study, human bones (n = 3) scanned using computed tomography were reconstructed as virtual 3D models (n = 6), and 3D printed using six commercially available printers, with osteometric data recorded at each stage. Virtual models and 3D prints were on average accurate to the source bones, with mean differences from −0.4 to 1.2 mm (−0.4% to 12.0%). Interobserver differences ranged from −5.1 to 0.7 mm (−5.3% to 0.7%). Reconstruction and modeling parameters influenced accuracy, and prints produced using selective laser sintering (SLS) were most consistently accurate. This preliminary investigation into virtual modeling and 3D printer capability provides a novel insight into the accuracy of 3D printing osteological samples and begins to establish an evidence‐base for validating 3D printed bones as demonstrative evidence.

ABSTRACT: This research investigates the metrology of 3D modelling and additive manufacturing (3D printing) osteological samples from computed tomography (CT) scans. Three bones were CT scanned, virtually 3D modelled and 3D printed using two printing methods, SLS and FDM. The virtual and 3D printed models were measured and compared to the original bone. The virtual models and 3D prints were found to be on average, acceptably close to the original bone (within 2.0 mm), but with variation between measurements and printers. This research provides initial data towards validating additive manufacturing in forensic anthropology, and the generation of an evidence base for grounding the technique in the forensic science process. The full publication from this research is available open access from the following link: https://doi.org/10.1111/1556-4029.13917

Virtual (or computer) 3D modelling of osteological material has facilitated the production of high impact images with countless applications, including the demonstration of evidence in a court of law. This research investigated the metrology and methods behind producing 3D printed osteological models, in order to validate the process for use in court. Three dry bone samples were scanned using computed tomography (CT) and the data modelled using two open-source programs: 3D Slicer (for initial development), and Blender (for further refinement). Several stereolithic (STL) models of each sample were generated using different parameters and compared with dry skeletal measurements. It was identified that, 1) The choice of viewing program can alter the 'appearance' of a model. 2) Both auto-smoothing in 3D Slicer and secondary smoothing in Blender, aid production of a model with less 'stepped' edges. 3) Fine/fragile bony elements (e.g. sutures or orbital bones) are not always fully included in the initial CT scan data/model. 4) Adjusting the 'label' (threshold map) in 3D Slicer can aid inclusion of these. The virtual models produced had an acceptable average observable difference (<1 mm to the dry material), but with larger varying ranges. Measurement precision and observer error varied with the methods, bones and landmarks used. These findings indicate further analysis into the effect of label maps, smoothing factors and viewing platforms is needed. The generation of accurate virtual 3D models relies on parameters being consciously selected based on empirically tested analysis. It is also suggested that guidelines must be created for valid use of 3D modelling in the presentation of evidence in court.