The method

How VeX Model Creator turns two scans into a model

The input is a jaw scan and a teeth scan. The output is the gum surface with the crowns removed and the root pockets left beneath it. What happens in between is five steps, none of which involve drawing new anatomy.

1. Two scans in

A jaw scan, which contains the gum and the crowns, and a teeth scan, which contains the tooth anatomy including the roots. Over the crowns the two coincide, because they are records of the same surfaces. That coincidence is the whole basis of the method. Scan requirements ›

2. Finding which way is up

Scans arrive in whatever orientation the scanner and the technician left them in. The occlusal axis is derived from the geometry itself, the model is rotated so the cut can be computed consistently, and the result is rotated back at the end. Nothing has to be aligned by hand.

3. Finding the gum line

This is the step that decides whether the model is any good, and it is where a height-based approach fails.

The gum line is not at a constant height. It scallops up between the teeth and down around each tooth neck, and the palatal vault sits higher than most of it. Cutting at a height would slice through the papillae and leave holes in the gum.

Instead, every point on the jaw scan is measured against its distance to the teeth scan. Over the crowns that distance is effectively zero — the two scans are the same surface. Below the gum margin the scans separate. The gum line is the level set where that separation crosses a threshold, which makes it the anatomical curve itself rather than an approximation of it.

The threshold is measured per case, not fixed. How well a given pair of scans is registered against each other varies, and a level that is correct for one pair can cut inside another pair’s noise and leave crown standing. The separation is measured first and the cut level follows from it.

The cut is then made exactly on that level set, splitting triangles where the boundary crosses them, so the edge is a smooth curve and not a staircase of whichever mesh edges happened to be nearby.

4. Removing the crowns, keeping the roots

Above the boundary, jaw surface that is also close to the teeth scan is crown, and it goes. The proximity condition matters: height alone would also delete the palatal vault, the vestibular ridges and the interdental papillae, all of which sit above the nearest gum-line point and none of which are crown.

Below the boundary, the socket walls come from the teeth scan — the real root anatomy that was in the patient’s mouth. Nothing is synthesised to fill the socket. This is the difference between a VeX and a scan with the teeth deleted: deleting leaves a hole that has to be filled with invented surface, which is no longer a record of the patient.

The root surfaces are also turned to face outward from the gum material rather than outward from the tooth. They are the walls of a cavity now, not the outside of a solid, and CAD software that culls back faces would otherwise render every socket as a hole.

5. Closing the model

Gum and root are welded together along the neck of each tooth, so the finished model has one boundary — the outer rim — and no daylight through the seams. Small trim slivers and genuine gaps in the gum are closed; the tooth channels themselves are not, because they are supposed to be open.

The result exports as a binary STL, ready to load alongside the rest of the case.

What it deliberately does not do

It does not smooth or beautify the surface. Eroding edges, filling openings and moving vertices are the three things a clinical model must not have done to it.
It does not add anatomy. The only new geometry is the thin seam closing gum to root.
It does not guess at a missing root. If the root data is in the scans, it is used; the model is a subset of what you supplied, not an interpretation of it.

Frequently asked questions

How does the software know where the gum line is?

It measures, for every point on the jaw scan, the distance to the teeth scan. Over the crowns the two scans coincide; below the gum margin they separate. The gum line is the level set where that separation crosses a threshold, so it follows the real anatomical curve rather than a horizontal height.

Does it work if my scans are not oriented with the teeth pointing up?

Yes. The occlusal axis is derived from the scans themselves and the model is rotated back to the original orientation afterwards, so any orientation is accepted.

Does the software invent any geometry?

No, beyond the thin seam that closes the gum surface to the root surface so the model is watertight. The root pockets come from the teeth scan and the gum comes from the jaw scan. Nothing is synthesised to fill a socket.

How long does one model take?

Generation runs in minutes. Against preparing the model by hand in CAD, beta testing has averaged around thirty minutes saved per model.

What happens to the crowns?

They are removed. A jaw triangle is only treated as crown if it sits above the gum-line margin and is also close to the teeth scan — the proximity test stops the palatal vault and the papillae being trimmed away as if they were crown.

Try it on a case you already know.

The fastest way to judge the cut is to run a case whose correct answer you already have in your head.

Create a VeX model ↗