| Developers: | PNIPU Perm National Research Polytechnic University |
| Date of the premiere of the system: | 2024/12/05 |
| Branches: | Pharmaceuticals, Medicine, Healthcare |
| Technology: | Robotics |
The main articles are:
2024: Developing a system for qualitative evaluation of dental treatment
Scientists at the Perm Polytechnic University have developed a system for qualitative assessment of dental treatment using a robot simulator. The university announced this on December 5, 2024.
Earlier, scientists from the Perm Polytechnic University developed an anthropomorphic robot simulator - a simulator for dental students with AI technologies. Neural networks, firstly, allow you to conduct a dialogue with a robot, and secondly, they are needed to recognize objects in the image in order to evaluate the results of dental treatment by students. At the same time, the system should reliably localize and evaluate in detail the changing object itself - the tooth in the oral cavity of the simulator, its properties and how they change during the operation. To do this, PNIPU scientists developed a two-stage recognition scheme and improved processing methods, which increased accuracy by up to 92% in unstable shooting conditions. Now the neural network evaluates not only quantitative indicators (dimensions, depth of the hole for the seal, thickness of the removed enamel layer), but also qualitative ones, for example, whether milling is done correctly, whether there are bevels, whether the bottom and walls of the tooth are uniform.
The study was conducted with the financial support of the Perm REC "Rational Subsoil Use." The "anthropomorphic dental simulator" project is a simulator for dental students where students can safely work out their skills in carrying out basic procedures - caries treatment, crown tooth treatment, canal removal and treatment. The built-in neural network using video cameras allows you to evaluate the results of work by processing the resulting images.
Modern neural networks are able to identify many objects of different classes without the use of any additional schemes. Usually, a simple one-stop neural network is used to search and classify objects in photographs. For example, it can find teeth in the jaw of the simulator with high accuracy, despite the constant change in the illumination and the shape of the object itself during treatment. But if it is necessary to analyze not the object itself, but only its part, for example, a small seal, the task becomes more complicated, the number of false positives increases. The neural network may mistake glare and irregularities inside the oral cavity for the desired holes in the tooth or completely miss them.
Scientists of the Perm Polytechnic University have developed a two-stage recognition scheme that analyzes photos in search of compound objects (individual teeth), cuts, normalizes them in size and analyzes each fragment separately to determine the desired small objects (seals, holes).
 | At the first stage, the area of interest is searched, i.e. the first neural network determines only the objects "tooth" and "tooth with a hole." They are cut out and transferred to the second stage, where the holes in the teeth and their properties are already recognized, - explained Andrei Kokoulin, associate professor of the Department of Automation and Telemechanics at PNIPU, candidate of technical sciences. |  |
Photo preprocessing is especially relevant for determining the properties of small objects, since their changes are more difficult to detect. It eliminates noise, increases contrast and brightness, and improves clarity, making the image more informative. Due to the fact that the teeth have a color close to white, the contours of the cut holes are poorly visible on them. Also hindered by the illumination necessary for the operation of cameras. Polytechnics have further incorporated into the system a contrast improvement program that preserves local details and image structures, which is important for accurately determining the boundaries of small objects in an image.
The shape of the tooth is a curve, and during the procedure it is important to calculate the size of its borders, the depth of the holes and the amount of removed enamel. To do this, scientists have developed a method for measuring an object of complex shape, allowing calculations in three dimensions.
| | The application of our two-stage system to 92% increased accuracy and to 5% reduced the number of false positive positives. For each treatment option, the neural network can determine its quantitative parameters. For "caries" and "canal" - the dimensions of the cavity under the seal, for "crown" - the thickness and uniformity of the removed layer on the sides and top of the tooth. As well as qualitative indicators - whether the treatment was correctly performed, whether the tooth broke during removal and how even the walls were, - shared Andrei Kokoulin. | |
Polytechnics note that in the future it is possible to create a mobile application with which you can photograph a cured tooth (even without a filling or crown) and assess the quality of treatment. Also, the proposed analysis method can be used wherever you need to survey various composite structures and mechanisms with many parts. The neural network-based system developed by PNIPU scientists significantly improves student education on a dental simulator, and also makes a great contribution to the development of modern technological medicine in Russia.
| The site content is translated by machine translation software powered by PROMT. The machine-translated articles are not always perfect and may contain errors in vocabulary, syntax or grammar. Read original article If you find inaccuracies or errors in the results of machine translation, please write to editor@tadviser.ru. We will make every effort to correct them as soon as possible. |