DICOM · STL · PLY · OBJ · 3MF · NIfTI · PNG / JPG / TIFF · MP4 / WEBM / MOV video. Read by content, parsed under a safety gate. Drop several at once. Click to select.
Analyses
§ 1 Identity — by dentition
Submit a query arch and one or more reference arches; the query is best-rigid-fit to each and the closest surface is the identity.
§ 2 Bone-level change — radiographic
Measured crestal recession between two timepoints (px). A "changed" finding requires the conformal interval to clear the clinical threshold τ; false progression is bounded by α.
§ 3 Surface change — 3D
Displacement (mm) against calibrated reconstruction noise. "Changed" fires only when the interval lies entirely past the change threshold.
Findings of record
No analysis on record yet. Submit a specimen and run an examination above.
Each finding is conformal: issued only when the interval around the measurement lies entirely past the threshold, so the false-alarm rate is bounded by α in finite samples — distribution-free, no model assumptions.
Method. Identity: best rigid fit (PCA-init + Generalized-ICP / learned point correspondence), smallest surface distance.
Bone-level change: reference-cancelled sub-pixel crestal registration. Surface change: de-biased regional displacement vs calibrated noise.
Every verdict carries a conformal interval with finite-sample false-alarm rate ≤ α. Provenance. All computation is local to this machine; no specimen leaves the device. This report records the inputs and parameters exactly as examined. Limitations. Validated on public single-timepoint data with synthetic perturbations; numbers are optimistic ceilings. Not a cleared medical device — advisory only, expert-confirmed.