Geomagic control

A scanning powder (METAL-POWDER Dry blue, R-dental Dentalerzeugnisse, Hamburg, Germany) was sprayed on the plaster models to avoid reflections from the metal surfaces. Every model was scanned to digitize the post-bonding bracket positions (TRIOS^®3W, 3Shape, Copenhagen, Denmark). Both pre- and post-bonding STL data were imported to Geomagic Control^® (3D Systems Inc., Rock Hill, SC, USA). Every tooth was cut out and saved both in pre- and post-bonding situation. In the image-processing software, the corresponding teeth were superimposed with a local best-fit alignment (Figure 5) and resulted in three linear and three angular measurements for each bracket.

The STL files of individually segmented teeth, including their root parts, were aligned with intraoral-scanned crowns to fabricate composite teeth. The maxillary and mandibular teeth were aligned in their position of maximum intercuspation. Subsequently, the crowns of the segmented teeth were removed and replaced by intraoral-scanned crowns. To establish a baseline for measuring root exposure from the alveolar bone, we trimmed the tooth portion above the alveolar crest from the maxilla and mandible STL files. The STL files were manipulated using Geomagic Control (version 2015; 3D Systems, Rock Hill, SC, USA) and Meshmixer (version 3.5.474; Autodesk, Mill Valley, CA, USA) (Fig. 2A–D). Meshmixer was used for trimming the teeth part from the alveolar bone and making the disc-shaped reference planes, and Geomagic Control was utilized otherwise.Figure 2

Setup process of the crown setup, root setup-1, and root setup-2. (A), The initial intraoral scan. (B), Teeth segmented from the initial CBCT. (C), Composite teeth of the intraoral-scanned crowns and CBCT roots. (D), Segmented maxilla and mandible with the tooth portion above the alveolar crest removed. (E), Crowns cropped from the initial scan. (F), Crown setup using the cropped crowns. (G), Root setup-1 using the composite teeth. (H), Root setup-2 considering the alveolar bone. CBCT, cone beam computed tomography.

Two geometries were used to assess dimensional accuracy and precision of nylon and PCL structures. The first was a 10×10 mm cube with 1×1 mm rectangular studs on each face. The 10mm and 1mm features were measured with calipers for 10 samples produced over 3 sintering runs. The second geometry was the full (nylon) or reduced (PCL) diamond lattice model. Lattices were μCT scanned as described above, reconstructed, and exported as STL files. 3D scans were aligned with their corresponding models using Blender (blender.org) and Geomagic Control (3D Systems, Rock Hill, SC). Aligned scans and models were divided into corresponding 100μm slices using Creation Workshop (envisionlabs.net). A custom MATLAB script was used to quantify overlapping and non-overlapping pixels in each slice for each scan/model pair.

We used Geomagic Control (2014 © 3D Systems) to compare the 3D reconstructions from 2014 to 2015. We aligned the two models, deleted all unnecessary data (e.g. substrate around each colony), and manually closed each 3D reconstruction by filling any small holes on the 3D reconstruction, which were present if occlusions occurred during the imaging process. For instance, it is not possible to photograph the base of the stalk of a tabular coral, given that it is attached to the reef substrate, so this hole was filled using a flat plane. Within Geomagic Control, model alignment was performed using the N-point alignment first, followed by the best-fit alignment functions. Once aligned, the coral colony was selected using the lasso tool and unnecessary data was deleted. Then, the 3D reconstruction was closed manually using the fill single function. Reconstructions of the same colony across time steps were compared using the 3D compare function, and metrics for volume and surface area were quantified using the analyses tools.

For every patient and each design version (V1, V2), two STL-files were imported into a 3D inspection software (Geomagic^® Control, 3D Systems Inc., Rock Hill, SC, USA), one showing the digital model with the virtually planned bracket and tube positions, the other showing the actual attachment positions after performing IDB. The individual tooth surfaces of the corresponding pre- and postbonding STL-files, which served as reference structures, were separated from the respective dental arches and superimposed using a measurement algorithm specifically scripted for this purpose, first described by Koch et al. [27 (link)] conducting an automated and successive best-fit alignment (Figure 5). Since coordinate systems were virtually integrated into every attachment beforehand, the deviations between their respective centers after superimposition were considered equivalent to the difference between the planned and actual attachment positions on the respective vestibular tooth surfaces. Using this procedure, the deviations were quantified in the reference coordinate systems in three linear values along the axes (mesiodistal (X, mm), vertical (Y, mm), orovestibular (Z, mm)) and three angular values around the axes (torque (X, degree), rotation (Y, degree), tip (Z, degree)).
The complete workflow is shown in Figure 1.