Artificial intelligence supports dentists in analysing X-ray images.

Whether it be for the prevention of future pandemics or for organ transplantation, Germany's doctors agree that significant advances in medicine can be achieved with the help of digitisation. Hospital doctors in particular believe that digitalisation offers primary opportunities for the healthcare system: 86 per cent emphasize the possibilities, while the figure for practice doctors is somewhat lower at 53 per cent - this according to a survey conducted in November 2020 by the digital association Bitkom in collaboration with the doctor's association Hartmannbund, in which more than 500 doctors in Germany participated.

"Dentistry is characterised by its high affinity for technology and stands out as a pioneer of digitisation in the healthcare sector," says Felix Goldschmid, CEO of dentalXrai, from his practice. “Today, treatment planning and implementation are often done with the aid of computers. New digital applications such as 3D diagnostics, 3D printers and intraoral scanners have decisively shaped the field of dentistry in recent years. Many of the innovations are now regarded as perfectly normal in dentistry today.”

The product launched by his start-up dentalXr.ai (derived from X-ray and AI – Artificial Intelligence) is also on the right track. After all, the software makes the dentists' job easier and allows them to spend more time with the patient.

This is made possible by a sophisticated system that relies on AI, in particular machine learning. "DentalXrai has trained neural networks with many thousands of dental X-ray images and has thus learned to autonomously detect previous treatments such as fillings, crowns and implants as well as pathologies such as caries, periodontal bone loss and apical lesions on X-ray images," explains Goldschmid, who took over the role of CEO from co-founder Hans-Peter Karpenstein in October 2020.

"The resulting artificial intelligence is now able to assist dentists in providing diagnoses as the software creates a digitalised dental chart of the patient in a structured format that highlights any relevant detections on the X-ray image in colour.” The findings are then stored in our own software. "The great thing for the patients is that the colour presentation of the detections means that even laypersons can understand the state of their health," he says, citing a further advantage.

AI finds three times more caries

A study has shown that this “digital second opinion” is safer and even more reliable than a diagnosis conducted by human. "The sensitivity, i.e. the proportion of caries found, increases when using our AI - in fact significantly," says dentalXrai's CEO. “In some cases, we found three times more caries, especially for early caries, when AI was used. This is significant because early caries detection is the first step towards early, less invasive treatment. AI helps to find caries in good time and enable painlessly treatment. This is also relevant for the cost of treatment to the patient.” The reliability of the software is largely due to the intensive training carried out by the current 41-person team of doctors, data scientists and software developers at dentalXrai.

The underlying models were and are optimised with the help of a vast dental data set. “We used X-ray images from Charité and dental cooperation partners all over the world. Dentists previously used to have to mark pathological changes or traces of previous treatments on these X-ray images,” says Prof Dr Falk Schwendicke, Head of the Department of Oral Diagnostics, Digital Dentistry and Health Services Research at Charité Medical University Berlin and co-founder of dentalXrai in an interview with Braincity Berlin

At least 5,000 images have to be processed for each detection in order to create an initial, robust algorithm, adds CEO Goldschmid. "The software we feed identifies statistical patterns from the data pool," says Schwendicke, who works for the WHO/ITU Focus Group on Artificial Intelligence for Health. "It is very important that our software does not take charge of the dental examination and does not decide on treatments," he emphasizes. "But it speeds up the analysis of X-ray images enormously and raises them to a high-quality, standardised level."

Focus on speed

The software detects caries, infections, but also implants and root fillings within just five seconds, but it can take twice as long for more complex data. For the user, the waiting times for both are hardly noticeable, says Goldschmid. “Speed has been a big focus for us – from the start,” he adds. It is precisely the viewing and assessment of X-ray images that take up a lot of time in dental practices and puts dentists' patience to the test. Not to mention that of the patients. Since 2017, a team from Charité in Berlin has therefore been working on the question of how machine learning can make things easier.  

“The idea originally came from other medical fields. I had read publications about the success of using artificial intelligence in the field of melanoma classification and thought to myself: How can AI be used in dentistry in a meaningful way?” explains Schwendicke, who as a member of the World Dental Federation has an impressive network.

Together with his colleagues, the oral and maxillofacial surgeon Dr Robert André Gaudin and the geoscientist Dr Joachim Krois, who worked at the Charité Medical University Berlin as a data expert specialising in deep learning and AI, he was able to get the pilot off the ground. Shortly thereafter, the team applied for funding through the BIH Digital Health Accelerator (DHA), a joint technology transfer programme by the Berlin Institute of Health and Charité. For two years, the project was supported by grants, co-working spaces and the programme’s experienced consultants. In 2020, the team was ready for lift-off: The company dentalXrai was founded - Charité's first dental start-up.

Launch with seven-figure turnover

Dentists have been able to take advantage of the new possibilities for detection since February 2021. Having the software, which is a medically approved product, integrated into synMedico's infoskop system practice communication platform from the start was "very important in order to enable dentists to use it without hindrance", says Goldschmid. This means that dentists do not need any additional technology. A digital X-ray machine in the practice is sufficient.

"Using tablets from synMedico, dentalXrai enables mobile and live AI-supported, automated X-ray diagnostics on the patient and offers immediate access to treatment planning and information about the patient," says the dentalXrai CEO. "This is welcome by dentists as well as patients.” The launch in 2021 was accordingly successful, which thanks to a seven-digit turnover, immediately made the Berlin company one of the leading AI dental companies in Europe.

But the team does not want to rest on its laurels: "2022 is an extremely important year, because we want to strengthen the rapid growth in Germany and Austria and successfully launch in other European countries through planned internationalisation for 2022," says Goldschmid. "We are currently looking for reinforcements in the areas of sales and customer management to complement our strong, technical team of IT specialists and dentists."

Even if the dental workforce is scattered all over Germany and on-site staff have become smaller in times of WFH, the young company finds its location in Berlin the ideal place to recruit good employees. "In addition, an ecosystem of AI start-ups is emerging in Berlin, including on the AI campus in Berlin Mitte, which helps promote the sharing of ideas and experiences," says the CEO.

What applies to IT specialists applies even more to colleagues in the medical sector: "We now work together with clinics and institutes worldwide to build up the largest and most diverse data pool possible and to network with experts from the different countries," says the CEO. "However, many activities in the medical sector are still carried out in cooperation with Charité and many of our employees have previously worked or studied at Charité.

The proximity to Charité and the attractiveness of Berlin as a location provide an ideal environment for the start-up to maintain or even expand its pole position in the young field of AI-supported programmes in dentistry. "The application of AI-supported programs in dentistry has only just begun." Felix Goldschmid has a vision: "In five years, dentalXrai will be available globally on the most important markets and will make it easier for dentists to carry out diagnostics and treatment planning using different types of imaging and to be digitally connected with their patients.”

And his colleague Schwendicke adds: “Dentistry has the advantage that as a field it picks up new trends several months later rather than sooner. We will not repeat mistakes made in other medical fields.”