Artificial intelligence in the fight against tooth decay

dentalXrai  is the first dental start-up to be spun off the Charité. It was launched via the accelerator of the Berlin Institute of Health (BIH). We spoke to one of its founders, Professor Dr Falk Schwendicke, Chair and Head of the Department of Oral Diagnostics, Digital Health and Health Services Research at Charité – Universitätsmedizin Berlin.

Dr Schwendicke, founding a start-up in the medtech sector is not easy even in a start-up capital like Brain City Berlin, is it? But you have done just that, creating a company that will develop an AI-based diagnostic tool for dentists. How did you come up with the idea? 

The idea originally came from other medical fields. I had read articles about the successful use of artificial intelligence in classifying melanomas, for example, and thought to myself: how might AI be made useful in dentistry? I spoke with my colleague Dr Joachim Krois, who works as a data expert in my working group at Charité. Together we got a pilot off the ground three years ago and launched the dentalXraiPro project. 

What can dentalXraiPro do?
 
The software we developed can use AI algorithms to automatically analyse dental x-ray images and highlight the findings in colour. It provides a pre-evaluated image in just a few seconds. In order to develop these algorithms, we had to train our software intensively. We used X-rays from the Charité and cooperation partners from around the world. The dentists marked pathological changes and traces of previous treatments on these x-ray images. We fed this pool of data into the software and it uses that to identify statistical patterns. It learned to distinguish between cavities, infections and root fillings, for example. 

What do dentists and what do patients get out of the software?

What's important is that our software does not eliminate the need for a dental examination and doesn't decide how to treat the issues it identifies. But it does speed up the analysis of X-ray images tremendously and raises it to a high-quality, standardized level. Analysing X-ray data usually takes a lot of time in the dental office. With dentalXraiPro, dentists can have a diagnosis in their hands with just a few clicks. This makes their work easier and they can take more time to talk to the patients. Patients can, in turn, be included in the diagnosis, because the coloured markings allow them to see the existing problems and better understand the proposed treatment. 

Don't you need an awful lot of data to train an AI sufficiently to ensure a reliable diagnostic result?

You need a lot of data, but not an infinite amount. The required quantity is more in the thousands range. Our focus is therefore more on a greater variety of data. We are concerned with the quality of the data used. How good and how clean are they? How many experts have looked at it before and above all: how generalisable is the data? These are also questions that we are addressing in the Artificial Intelligence in Dental Medicine (AIDM) working group founded in 2019 as part of the Deutsche Gesellschaft für Zahn-, Mund- und Kieferheilkunde (German Society for Dentistry, Oral and Maxillofacial Medicine ) DGZMK and the Artificial Intelligence for Health (AI4H) dental working group at ITU/WHO that Joachim Krois and I manage together.

There are already competing products on the market. What are the unique selling points of dentalXraiPro?

We differ from other companies in the field in three ways: first, we don't come from the tech field, but from the dentistry. This is reflected, for example, in the usability: our software can be used intuitively by dentists. Second, it is very transparent and based on evidence. After all, it has to meet our scientific standards. And thirdly, compared to other products, we have come a long way with regard to the approval of dentalXraiPro as a medical product. The certification of the products is a major difficulty in the medtech area. 

dentalXraiPro is an example of the successful transfer of medical knowledge into other areas. It was therefore also funded through the BIH Digital Health Accelerator (DHA), the joint technology transfer programme of the Berlin Institute of Health and the Charité. How did this support help you?

We applied for funding with our pilot and joined the accelerator in 2018. The support included funding, co-working space, and mentoring by experienced consultants on such mattes as legal questions and setting up a pitch deck. We were thinking entrepreneurially from the start, but as a start-up you still need specific advice on many issues. Most importantly, the accelerator enabled our team to grow organically. We are nine now. The team members have different areas of expertise: dentists, data scientists, and software developers. I especially bring the medical expertise to the enterprise because I am still working at the Charité. The Charité itself supported us with data, knowledge, and space. The funding from the BIH Accelerator has now expired.

To what extent does dentalXrai benefit from Berlin as a science location?

Berlin has an extremely strong ecosystem in the field of medicine. BIH is a part of that. Of course, we benefit from this, especially the quick access to knowledge and contacts. The city is also an attractive location for data and software experts. This makes it easier for us to recruit good employees. The close connection to the Charité is also an essential location advantage for us.  

What will happen to dentalXrai after leaving the BIH Accelerator?

A first milestone will be the market launch of our product at the beginning of next year. To do this, we first need certification as a medical product. The second step is to win over multipliers in the market, for example, manufacturers of X-ray machines and patient management systems. Our financing is secured for now.  

And where is the future headed in the field of AI-driven dentistry?

Exactly in the direction we are striving for: making better use of existing data. In the future, we will know much more about individual patients than we do today: through x-rays, photos, clinical and case histories, maybe even social media data. Data dentistry will also significantly improve prevention in dentistry and personalise diagnosis and treatment. Another advantage for dentistry is that it is a field that picks up new trends a few months later rather than earlier. We won't be repeating mistakes made in other medical fields. (vdo)