LIAO2018 ATTRIBUTION
Current source-family notice. This copy retains the bundled attribution, processing and reuse information. The licensing overview describes current positions and unresolved distribution permissions.
Liao 2018 microCT third-molar specimen Source: Che-Wei Liao, Lih-Jyh Fuh, Yen-Wen Shen, Heng-Li Huang, Chih-Wei Kuo, Ming-Tzu Tsai and Jui-Ting Hsu (2018), Self-assembled micro-computed tomography for dental education, PLOS ONE. https://doi.org/10.1371/journal.pone.0209698 Public S1 dataset: https://doi.org/10.1371/journal.pone.0209698.s001 License: Creative Commons Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/ One scanned maxillary right third molar supplies Universal #1. Universal #16 is an explicitly mirrored counterpart with reversed triangle winding, not another specimen. The Skyscan 2211 source comprises 517 reconstructed TIFF slices. The study reports 36 micrometre sampling; this is not a measured accuracy claim. ToothForm derived hard tissue from the largest intensity >60/255 component. Enamel and dentin are a core-seeded gradient watershed partition of that fixed mask, with dentin cores at least 15 voxels from air. A disconnected 44-voxel high-density root island was assigned to dentin. A further 53,530 existing partial-volume boundary voxels were classified as enamel only where they lay within 5 voxels of air and prior enamel and their nearest confident source-intensity tissue sample was enamel. This inferred boundary labelling changes no whole-tooth geometry; it is not an expert tissue annotation. Pulp is the largest connected axial enclosed-void component. Masks were converted to 0.5 isosurfaces after Gaussian smoothing at sigma 0.6 voxels. All layers share the same calibrated scale, rotation and translation. Meshoptimizer creates detail and atlas display tiers by retaining source vertices/normals. Native SceneKit archives preserve the exported geometry. Tissue colors are display colors; no photographic color data were supplied. The original volume and full-density isosurfaces remain in the research cache. Tiny tissue contacts at the first/last supplied slices create finite end caps. Accessory canals, apical foramina and disconnected spaces are not validated. Automatic segmentation was checked against source sections, but has not been compared with expert labels or clinically validated. No artificial canal connectors, idealized crown, nonuniform scaling or synthetic anatomy was added. Exterior detail/atlas display normals are a derivative: Gaussian sigma 0.055 mm, radius 0.15 mm, at most 64 neighbors, with area and normal-angle weights and a 25-degree maximum deviation from each stored source normal. Vertices, triangle indices and all non-normal OBJ records remain byte-identical. Mirrored normals are exactly reflected; pulp normals and all source geometry remain unchanged.