© 2023 MJH Life Sciences™ and Dental Products Report. All rights reserved.
AI-powered diagnostic systems can improve the effectiveness of identifying caries on radiographs.
Artificial intelligence (AI) is no longer a new concept in dentistry. It has been seamlessly integrated into numerous functions in dental practices, including practice management software, telehealth capabilities, image-guided oral and maxillofacial surgery, marketing campaign management, and now, perhaps most importantly, diagnostics and clinical decision-making.
“It’s clear that AI in dentistry, in every facet of running a dental practice and in health care, is going to continue to grow,” Albert Kim, CEO of SOTA Cloud, says. “The hope is it will continue to have a huge impact ultimately in patient outcome. Over the past 20 years, we’ve gone through a digital revolution in dentistry, and AI will compound that growth even faster and have a greater impact. With AI and cloud-based technology, we are in the early stages of unlocking the true potential of the digital revolution in dentistry.”
When it comes to diagnostics, AI can be effectively implemented in dental radiographs using neural networks.1 For instance, conventional neural networks (CNNs) quickly and precisely analyze images. This AI can detect tooth decay, restorations, periapical lesions, root fractures, and other anomalies in a fraction of the time it would take a human to do so—and oftentimes, more accurately.
A second type of neural networks are artificial neural networks (ANNs), which are programmed to detect patterns within data sets. This AI can be utilized to map anatomy of teeth or perform facial recognition. With both variations of AI, large amounts of data are fed into the systems so the technology can “learn” to identify these patterns and anomalies, growing more accurate as it is fed increasing amounts of data—data that would take humans extensive amounts of time to sort through.
“The power of AI comes from its ability to analyze vast amounts of data, identify patterns and trends, and automate tasks that used to be time-consuming for humans,” says Teresa A. Dolan, DDS, MPH, chief dental officer of Overjet. “AI is already becoming mainstream, and I predict most dental practices will adopt AI technology [once] they understand the benefit it brings to patient care.”
Accuracy and Efficacy
And the benefits are impressive, primarily stemming from the accuracy, reliability, and efficiency of the technology. A 2018 study examined the efficacy of CNNs at identifying dental caries on periapical radiographs. The results showed that CNNs had detection accuracy rates of 89% for premolars and 88% for molars.2 Even more impressive results came from a 2020 study that found that AI could identify caries with a 97.1% accuracy rate.3 Results from the second 2020 study determined that AI successfully detected 97.6% of dental restorations on x-rays, leaving no doubt that the technology is effective.4
“There’s a real human factor when it comes to a profession like dentistry because there’s a lot that can go on,” Kim says. “Even in a single periapical or bitewing x-ray, it is like the equivalent of Where’s Waldo, but the target isn’t specified.”
This, Kim says, is why there are so many missed potential treatments and cases. There are constantly numerous things happening at once and the dentist is beholden to an x-ray image that may not be up to par for the full diagnosis. “There’s a lot that can be missed with many tools up to this point in the history of dentistry, especially when the dentist doesn’t have the luxury of time [and has so] many patients to care for,” Kim explains. “With today’s AI technology, the dentist can have the best tools available to make the best diagnostic inference. We’re now seeing—and building—the utilities to help dentists give better care than ever before.”
To evaluate accuracy among human practitioners in comparison with AI, a recent study by Pearl had dentists and a computer vision/learning machine (CV/ML) examine a sample of 10,000 radiographs to identify caries.5 The study had the humans and AI score the x-rays for the absence or presence of caries. It examined whether the participants agreed or disagreed and how that compared with the AI results. Researchers found that 3 clinicians unanimously agreed on the presence of caries or decay only 4.2% of the time and that the AI was better at determining the existence of caries on the x-rays than the human evaluators.
“Computers are just better than us at many things, like math or crunching numbers,” Ophir Tanz, founder and CEO of Pearl, says. “But they’re worse at issuing care and having empathy and being able to take lots of disparate data sets and come up with a treatment plan, etc. I think that we should be embracing the technology to support us in things that it is uniquely positioned to enable us to do better.”
Dr Dolan agrees about the advantages of AI in regards to data analysis and consistency.
“AI helps overcome one of the biggest challenges in dentistry: inconsistency,” she says. “Dentists’ training, experience level, philosophy, and even fatigue can influence their diagnosis. Artificial intelligence is objective.”
A Second Set of Eyes
This objectivity can be a valuable tool in the clinician’s arsenal and a powerful diagnostic tool to complement a dentist’s capabilities and assessments. However, professionals say, it is in no way designed to replace the dentist.
“Machine learning and AI-powered applications provide dentists an objective view when faced with a difficult diagnosis,” says Lea Al Matny, DDS, MS, a clinical education specialist at Carestream Dental and an oral maxillofacial radiologist with SeeThru Reports. “This in no way undermines expertise, training, or the skilled diagnostic integrity dentists bring to analyzing x-rays and providing patient care. But it augments the capability, accuracy, and the precision of the diagnosis and treatment.”
Ultimately, it can never hurt to have a second set of eyes, particularly for a busy dentist trying to juggle all aspects of the practice without sacrificing patient care.
“No dental school is sending practitioners in the field proficient enough at identifying pathology,” Tanz says. “Dentists are expected to learn out on the job, and maybe you get pretty good at it, but you’re still missing a bunch of stuff. What AI is doing is surfacing a new level of truth. As a result of AI, you’re getting better, more comprehensive, more accurate, and more precise detection of pathology. You can refer back to the clinical trials and see computers are able to identify pathology in a way that is superior to humans. AI eliminates that variability and ultimately we can unburden practitioners and allow them to focus on things that matter.”
Tanz sees AI as such a supporting tool that Pearl named its AI platform Second Opinion®. Second Opinion uses generative adversarial network (GAN) preprocessors to enhance x-rays when the images aren’t top quality. Image preprocessors and machine-vision models combine to detect restorative, pathologic, and anatomical features on images. The software segments teeth to assess the anatomical structures and identify various pathologies, including caries and calculus, restorations, margin discrepancies, lesions, and periapical radiolucency. Pearl has FDA clearance to evaluate for these conditions in both bitewing and periapical x-rays in patients as young as 12 years. Beyond the United States, various international regulatory boards have cleared Pearl for market in more than 90 countries.
“AI enables us to drive the highest standard of care,” Tanz says. “We’re just delivering a level of truth that is totally transparent, that is totally verifiable. But [it] is also provably statistically quite good. And we should be embracing truth above all else because that is what’s going to lead to healthier patients.”
And healthier patients are the key.
“AI-powered applications will make dental care more efficient by freeing up the dental team from mundane tasks,” Dr Al Matny says. “The focus will also be on improving accessibility of health care for a wider population at a more affordable cost, eventually leading to preventive dentistry.
Focus on Integration and Accessibility
The dedication to better patient care extends across the industry—after all, AI is only as good as the benefits it brings to patients. One way to increase these benefits is to make AI more efficient and accessible. To achieve this, several industry partners are teaming up.
Cloud-based dental software company Planet DDS and dental-AI company Overjet recently announced a partnership to incorporate AI capabilities into Planet DDS’ Apteryx XVWeb Cloud Imaging software and its Denticon practice-management platform. This integration will allow clinicians to improve diagnostic accuracy with Overjet’s AI from any imaging device, thanks to the scalability of the Apteryx XVWeb software.
“Cloud and AI technologies are transforming the practice of dentistry,” Eric Giesecke, CEO of Planet DDS, said in a news release announcing the partnership.6 “With this integration, dentists can harness the power of both technologies to give patients more transparency and confidence in diagnosis and treatment planning.”6
This partnership is an added bonus to Overjet’s already-robust capabilities. Overjet is the only AI provider with FDA 510(k) clearance for bone level quantification (including measuring distal and mesial alveolar bone levels in bitewing and periapical radiograph), and its software gives clinicians simple quantification when measuring bone levels, eliminating the need to visually estimate the measurement from the cementoenamel junction to the crestal bone to determine if there is bone loss.
“Having the AI-powered quantification not only helps providers, but also can provide the radiographic evidence of bone loss that most dental payers require in adjudicating periodontal treatment claims,” Dr Dolan says. “Dentists have the information needed to help diagnose both restorative and periodontal treatment needs.”
To gather the information necessary to support this AI, Overjet’s machine-learning algorithms have processed more than 28 million images from dental practice clients. Fifty licensed dentists annotate images from these clients to continually train, test, and validate the AI models. According to an internal study, Overjet analyzed more than 7000 tooth surfaces and found that when dentists used the AI, they were able to detect 32% more tooth surfaces containing caries.
“Dentists want to provide the best care possible to patients,” Dr Dolan says. “Just as they may call over a clinical colleague to take a look at an x-ray and confirm what they see, they can depend on Overjet’s AI to provide an objective analysis. It draws the eye to potential areas of concern, which the dentist can then review and diagnose. AI elevates the dentist’s ability to diagnose accurately and consistently; it doesn’t replace it.”
The company doesn’t plan to stop developing and growing any time soon. The easy-to-use platform offers real-time AI analysis, and clinical and operational insights, and integrates with the digital imaging and practice management systems to draw data from prior visits to help identify potential treatment needs. This, Dr Dolan says, is just the beginning.
“Overjet is developing more AI models to detect pathologies in permanent and primary dentition,” she says. “It’s also exploring point-of-care adjudication with insurance carriers, which would eliminate one of the biggest concerns patients have: wondering if insurance will cover treatment. It’s working with universities to introduce dental students and residents to AI-assisted dentistry. And it’s evaluating ways to integrate dental and medical records to use predictive analytics to promote personalized health care recommendations that will improve patient outcomes.”
Overjet isn’t the only one committed to improving patient outcomes through collaboration. Pearl’s Second Opinion can now be easily accessed through SOTA Cloud software.
“With our SOTA Cloud software—built with years of insight into dentists’ workflow structure—we created a platform where AI can be seamlessly used within their current workflow, which makes it easy to add into their processes,” Kim says. “With our new update, users can simply click on a button in our SOTA Cloud software to access Pearl’s AI tool Second Opinion, so it doesn’t take any additional training or windows that need to be managed. Using AI just becomes second nature, like taking an x-ray and adding a filter. With SOTA Cloud, we are decluttering the workspace to help dentists optimize workflow and utilize AI at higher rates.”
To utilize AI effectively, Kim says, it all starts from the ground up by taking the x-ray and getting the raw data. That’s what feeds into the imaging software, such as Pearl’s AI.
“We are starting from the source of that data by applying our own AI in the form of enhancing the image so that it’s optimized so that when it goes into AI diagnostics, they’re putting in good high-quality data,” he says. “We’re embedding AI into our products so this new technology has the best workflow for adoption. We also put AI into our hardware (x-ray sensors) so we create the most optimized data, not only to enhance doctors’ visual diagnostics but to help other AI technologies like Pearl by feeding it good data.”
This integration is key for streamlining workflows. To do this, companies such as Carestream Dental are integrating AI diagnostic tools directly into their imaging software and clinical evaluation workflows. Once the image is acquired and displayed, an AI reading is available with a single click. The AI software automatically detects a broad range of pathological and nonpathological conditions on the x-rays and highlights findings in color, both on the image and on the interactive chart. Findings are automatically documented into a comprehensive report, which includes the x-ray image, a dental chart, and the list of findings, saved into the patient file. Carestream also has integrated AI into its CS 9600 CBCT system, which uses AI-powered positioning to ensure consistent results.
“AI applications can be used to reinforce confidence and increase consistency by using a real-time second opinion,” Dr Al Matny says. “The focus should be on increasing consistency, and as a result, improving patient health.”
An important step to improving patient health is securing patient treatment acceptance, which can be an onerous task. Patients may not understand why they need treatment or what markers are showing up in their current oral-health status that could develop into bigger issues down the road. One benefit of AI is its opportunity for patient education.
“Most patients don’t know how to read traditional x-rays, but when they see the AI markings that outline decay and measure bone loss, there’s an ‘A-ha!’ moment,” Dr Dolan explains. “Seeing the objective findings builds trust and confidence in their doctor’s diagnosis and treatment recommendations. Dentists and dental organizations have been enthusiastic about integrating AI technology into their practices, especially after seeing the value it brings to patient care and practice efficiency.”
Results of a 2022 study found that when patients were shown a radiograph with an arrow pointing to a lesion, they were much less likely to be able to see or understand it than when using AI-generated colored overlays highlighting the issue.7 Patients responded well to the AI-supported diagnostics and radiographic findings, thanks to the easily understandable visual explanation of the pathologies.
“Overjet’s AI instantly analyzes x-rays and applies color and measurements to draw the eye and help with clinical decision-making,” Dr Dolan says. “This AI does more than just help dentists diagnose accurately; it also gives them a powerful tool for patient education and communication.”
For all the benefits AI is bringing to diagnostics, there are many more coming down the road. Neural networks and AI are already excelling at detecting tooth decay, assisting with root canal system anatomy, performing cephalometric points marking in orthodontics, and more. To take AI to the next level however, dentists need to embrace what’s already there.
“There’s always a fear of adopting new technology because we are inherently not creatures that adopt change. Especially with something nascent like AI, it can be scary for a lot of doctors at first,” Kim says. “The less they know about it, the more they see AI as something to fear. But I think it’s important for them to understand this is really a technology that aids the process of the dentist who is ultimately the professional, [with] the dentist knowing they have the ultimate say and ultimate executive decision-making power, and that AI is just an aid.
“AI is just at the forefront, in its infancy,” Kim continues. “It’s going to require even more diligent surveillance of how this is affecting users and how this is ultimately bettering clinical care for the dental practice.”
Dr Dolan agrees that the industry is on the brink of something great. “We are just getting started,” she says. “AI gives dentists a powerful tool to help diagnose accurately and consistently, and we’re proud to lead the movement to make dental care more patient-focused.”
Ultimately, the consensus is that tools such as AI are only going to become more valuable over time, enabling dentists to provide better care and allowing them to focus more on the bigger picture.
“Dentists do a lot, and I think to the extent that we can unburden them so that they can more effectively issue care and do that in a way that that has benefits to patients, that’s what’s going to enable them to be better practitioners,” Tanz concludes. “And that’s a no-brainer.”