Vertebral model production

With institutional IRB approval, a CT (computed tomography) scan of an adult patient with 1-millimeter (mm) slice thickness was acquired from an anonymized, encrypted institutional database which does not record identifiable patient information. Patient anonymized DICOM (Digital Information and Communications in Medicine) data is deposited into the database after signed consent, and unable to be linked to identifiable information. The completely anonymized DICOM files were downloaded onto an encrypted hard drive and reviewed for inclusion criteria. CT reviewing was performed by three individuals (WC, AD, KF) on 3D Slicer (Slicer, v. 4.10.2, 2018). Criteria for inclusion of DICOM data were: inclusion of a complete vertebral column in the study, no surgical spinal instrumentation present on the CT, and no traumatic or iatrogenic deformity of the native anatomy (including previous laminectomy or disruption of posterior elements). Slice-based thresholding was then applied to the CT DICOM files with a range of 193–3000 Hounsfield Units. This particular range established the boundaries of the desired vertebrae in this particular DICOM data set. The vertebral interfaces were manually segmented in each individual CT slice in order to establish maximum accuracy of the vertebral bony associations. The thresholding and segmentation processes were performed by two individuals (WC and AD) with equally divided data sets, and quality and inclusion of thresholded anatomical structures was inspected on each slice by the opposite individual. Both individuals have extensive experience in thresholding and segmentation of anatomical structures using 3D Slicer, and independent assessment of data sets was performed to minimize observer bias. This workflow in combination with 1 mm DICOM CT slice thickness has been shown to be a highly accurate means of recreating spinal anatomic features with 3D printing⁴⁶. The finalized selections were inspected in a three-dimensional projection within 3D Slicer, rendered to STL (Standard Tessellation Language) format, and then edited for manifold assurance using Meshmixer (Autodesk, 2017). Cura (v.4.0, Ultimaker, Netherlands) software was used for slicing and production of the models. The STL files are loaded onto the virtual software platform in Cura and spatially arranged for maximum printer efficiency. A 1 mm outer shell with 0% infill is used to produce hollow models that can be filled with the polymeric foam. We chose a 1 mm shell in order to replicate the average thickness range of cortical bone in the human vertebrae⁴⁴. ABS filament at a diameter of 2.85 ± 0.05 mm is printed through a 0.4 mm nozzle at a bed temperature of 80 °C, nozzle temperature 250 °C, and nozzle extrusion speed of 70 mm/sec. PLA filament at a diameter of 2.85 ± 0.05 mm is printed through a 0.4 mm nozzle at a bed temperature of 80 °C, nozzle temperature 200 °C, and nozzle extrusion speed of 70 mm/sec. ABS and PLA cost expenditure is approximately $0.02 per gram or $0.46 per meter.