Straumann - AI gets on your nerves! Thank goodness!
Artificial intelligence is all over the news, and seems to be pervading everything. Our dental and periodontal professions must surely be looking at how it will affect oral health.
In this series of interviews, the European Federation of Periodontology is exploring what our partners are doing, or perhaps NOT doing with AI.
In this interview, we talk with Thomas Friese, chief technology officer of Straumann group.
Thomas Freise is adamant that AI is a big game changer in the world of implantology and Straumann is right up there with it. “We use AI throughout our portfolio” says Friese. “We have a long digital history also with coDiagnostiX as one of the leading surgical planning systems in implantology. A few years back we acquired Promoton, an AI start up in the dental space.” This has allowed Straumann to develop AI models that it can use specifically to dentistry, integrating it into Straumann’s existing software.
This all sounds great, but what, in practical terms, does this actually mean?
One very useful application is “segmentation”. Put simply when planning an implant, you have to check where all the nerves in the jaw are. This means analysing scans and mapping the nerve channels. For a human this is very laborious and time-consuming work. Not for AI though! “If you have to segment out the nerves that go through your lower jaw you need to follow the nerve channel painstakingly slice by slice”, explains Friese. “What you want to avoid is if you place an implant that you drill into that nerve it will essentially disable half of the face!”
That isn’t good! And that is where AI can quickly and accurately map the jaw by analysing the scans. Straumann also uses AI to do virtual tooth extractions, from which implantologists can accurately analyse the shape of the hole or socket left behind. “We have deep learning based AI models that perform segmentation of intraoral optical scans and radiological imaging [as the second big modality.]”
This is all excellent, but Thomas Friese says even that isn’t quite enough. He adds: “What you then need to do is identify the shapes of the roots of the teeth, or the bone versus the soft tissue”. Deep learning AI allows you to do this by segmenting the whole area into its individual components.
Put very simply, segmentation distinguishes between what is tooth and what is NOT tooth! This does not replace the professionals at all though. Friese explains: “It helps automate cognitively burdensome tasks. AI is there to augment the professionals; it is not likely that AI will replace dental professionals but dental professionals who use AI will replace those that don’t!” The other great thing according to Friese, is that AI is helping all the different professionals in the industry to work together. “We have a partner integration SmileCloud that is geared to specifically aid the collaboration of peer teams around a patient case, and those tools also rely heavily on AI.”
Straumann is now publishing AI models that can literally take a photograph of you smiling and then guess what the 3D structure of the mouth is. You can then design a smile from that selfie! So Friese believes that AI “aids the collaboration of care teams around a patient case.”
Friese is also optimistic about Straumann’s AI future, including its new “digital platform” AXS that it is working on that will “integrate various applications and data sources and ensures seamless data flow.” This will “allow professionals to collaborate via the platform and to exchange data,” he adds.
One practical use might be that dentists put various bits of information onto this platform, allowing Straumann’s planners to work out a treatment plan for surgery or whatever. That can then be shared and any changes or amendments can be made before actually going ahead. Moreover, different parts of the platform can be shared with different stakeholders, from dentists to patients to Straumann planners.
A digital one stop shop that should get rid of endless emails, phone calls etc…and one that both dentist and patient can use. It is already working in North America. Are there any dangers with AI? Not if you impose the same rigour as you would with other technology used in medical devices, believes Friese. “There is testing and clear scoring on how well AI performs on real world data. So hallucination is typically not a problem as it can be with AI like Chat GPT.”
Friese does add though that his biggest worry is that “the industry gets reckless in its application, but as long as we apply the very same quality standards that we do to every medical device then I don’t have any fear.”
That should settle a few nerves!