Bern Tooth007 mandibular canine — standalone demo 1107

Source: David Haberthür, Ruslan Hlushchuk and Thomas Gerhard Wolf (2021).
Automated segmentation and description of the internal morphology of human
permanent teeth by means of micro-CT. BMC Oral Health 21, 185.
https://doi.org/10.1186/s12903-021-01551-x
Dataset: ZMK Tooth Cohort, https://osf.io/rt3da/
Public source file: https://osf.io/7trgk/

The OSF project metadata has no dataset license. Article CC BY 4.0 terms and
its data-waiver statement do not independently confirm redistribution rights
for this external volume. This is a local attributed research prototype;
external dataset redistribution permission remains unconfirmed.

Tooth007 is an intact mandibular canine with undocumented source side. It is
not assigned a Universal tooth number and has no mirrored atlas counterpart.
The original 3,002 PNG slices are at nominal 10 micrometre sampling. A bounded
crop and exact rounded 2x2x2 block means create the 20 micrometre working volume.
Original data remain preserved. Sampling is not a claim of anatomical accuracy.

ToothForm derives hard tissue from intensity >30 and the largest
connected component. Enamel/dentin use a source-gradient watershed with deep
source-intensity core seeds. Ambiguous existing outer-boundary labels are
classified from their nearest confident tissue sample within 0.18 mm of air
and enamel. This changes labels within measured tissue only. Tiny final enamel
boundary islands are retained. Pulp is the largest connected enclosed-void
component. All tissues use one shared calibrated scale and proper rigid frame.
Isosurfaces use Gaussian sigma 0.6 voxels. Bounded native detail simplification
retains source positions/normals. The displayed pulp is a meshoptimizer
simplification to 149,908 triangles of the 738,370-triangle dense pulp
(2026-10-11); the dense source is kept. Compiled display archives carry baked
optical attributes and smoothed display normals; vertex positions are unchanged.

Ring/pepper scan artifacts and uncertain tissue interfaces remain. Small
disconnected spaces and apical/accessory anatomy are not validated. These
automatic labels are visually reviewed prototypes, not expert ground truth or
clinically validated anatomy. Tissue colors are display colors.