What the services do
A dental chain in Russia is piloting an AI module that analyzes the whole cone-beam CT (CBCT) volume. It marks caries, periodontitis, impacted teeth and traces of earlier treatment, and builds a base for a treatment plan (April 2026). A Russian image analysis service for dentists claims more than 60 conditions on 3D images and over 40 on 2D images (checked 29 September 2026). A telemedicine service checks photos of teeth for caries in 5–10 seconds, including hidden caries in the form of a white spot (August 2026).
Experience around the world
A developer of a pediatric algorithm in the USA reports an 80% improvement in correct detection of caries in children; this is a company claim, not a peer-reviewed study (February 2024). A network of more than 400 practices and 900 dentists rolled out AI image analysis across the whole group but has not yet disclosed any figures (July 2024).
Why it is a medical device
The deciding criterion is the purpose: if software interprets an image or assesses the risk of disease, it falls under the definition of a medical device (Article 38 of 323-FZ, Russia’s health law) and is registered under Government Resolution 1684. Under the national classification, software as a medical device with AI belongs to risk class 3 (Ministry of Health Order 4n, Annex 2, item 15.1.1). Under EAEU rules, diagnostic software usually falls into class 2b or 3. In August 2026, Roszdravnadzor, the Russian health regulator, reported 59 domestic AI medical devices registered in Russia. A visit counter, reminders and booking do not count as medical devices.
What this means for a clinic
When you choose an image analysis service, ask for the registration certificate number and the risk class. Do not offer patients a “photo check” without a registered algorithm and a license. Separate service AI, which you can roll out in 14 days, from medical AI, whose registration takes years, according to market players.