This text is part of the special research section
A team of researchers from Polytechnic Montreal is using artificial intelligence to reduce margins of error in the manufacture of dental crowns.
Hyperfaking in dental service, this is the motto of François Guibault, Full Professor in the Department of Computer Engineering and Software Engineering at Polytechnique Montréal. For 18 months, the researcher used the same artificial intelligence (AI) technology that is used to manipulate videos to make dental crowns.
“Tooth making is as much a science as it is an art,” he says. Each mouth has its history and each tooth is unique to the individual, by its shape, colour, surface and wear. If you take a tooth out of a catalog and put it in it, it will look artificial. »
The technician who makes the crown starts with digital measurements and photos and then, using various software, generates a three-dimensional shape that a machine then reproduces in porcelain. “We are looking for a way to further automate certain processes,” defines François Gibault.
a matter of etiquette
To create a tooth or any digital shape, we start from a cloud of digital points which we then connect into small triangles. The shape of the tooth is the sum of all these little triangles. “When you scan a tooth, you can have 40,000 points. It’s all about figuring out which one is better,” explains the professor.
Before being able to produce the figure, it is necessary to name each point, which consists in giving it a coordinate in x, y, z. In the case of the teeth, which sometimes touch each other, the system must also specify the address of each coordinate – in other words, which tooth it belongs to.
“It takes a lot of mouths to make it happen,” says François Gibault. He explains that the data comes from Kerenor Dental Lab, in Westmount, which contains thousands of scans and crowns. This collaboration also includes iMD Research, in Montreal, which specializes in dental technology.
“We now know how to classify each tooth with a reliability of over 98%. To achieve this, two students from Cégep Édouard-Montpetit, recruited by the College’s Center for Technology Transfer in Pharmaceutical Sciences, located in Cégep John-Abbott, worked for several months to analyze the point cloud and conduct Quality control and elimination of “noise”, i.e. irrelevant or redundant points.
But why create data on the mouth when only one tooth is needed? First, because teeth often lack parts that need to be replaced. But also because the tooth never works on its own. “We generate a tooth in its context. The ideal for us is to have two adjacent teeth and three opposite teeth.”
Make your neurons work
The next step is to create the shape to reproduce it. Therefore, it is necessary to reconnect all the points to create the decks one adds. François Gibault uses the so-called neural circuits. These are Generative Adversarial Networks (or GANs, for the English acronym), a deep learning technology developed in Montreal in 2014 that revolutionized artificial intelligence.
Deep learning consists of opposing two brain functions: those that generate and those that differentiate. The generator (also called the encoder) produces the shapes; The discriminator (called a decoder) determines whether it is similar to a model or not. The system learns by rejecting what is bad and preserving what is good.
In order for the system to learn, it must be provided with a lot of data. And this is where the data from the Kerenor dental lab is very important. We have reached several hundred mouths. The more we show it scans And the corresponding formats, the better. »
François Guibault hopes to launch the prototype by summer. For us, this will be the first opportunity to use and criticize it. This is the extent of genius. We strive for continuous improvement.”
The research engineer, who designed hydraulic turbine models before becoming interested in dental crowns, is already seeing other applications of his work. His lab uses the same techniques for the additive manufacturing, known as 3D printing, of titanium aircraft parts. But he also sees a benefit in the manufacture of prosthetics. “Instead of pulling your cane from the catalog, we can make your chin up for you.”
However, he insists that his current dental work is not aimed at depriving technicians with 20 years of experience from their livelihood, on the contrary.
“Yes, if we succeed, we will be able to create a shape in seconds, while a technician can do it in an hour using various software. But he assures that his art and experience will always take adjustments and his appreciation for the result. The goal is not to replace it, but to give it another tool that will shorten the intervention time and validation.”