Students hone their dental AI skills while helping their patients

A monochrome dental X-ray with different tooth surfaces highlighted in bright pastel colors

UF's College of Dentistry partners with dental AI company Overjet to teach students

After just one hour of training with Overjet’s dental AI technology, University of Florida dental students improved their ability to detect cavities by more than 40%. 

Since December 2023, 100 students have gone through the training on Overjet’s module. For many, it was their first look into the AI-powered future of their field.

“Students absolutely loved it. They were a hundred percent focused,” said Anita Gohel, B.D.S., Ph.D., the chair of oral and maxillofacial diagnostic services in UF’s College of Dentistry. Gohel has supervised dozens of students who trained on Overjet’s system and is researching how the application can serve as a dental AI teaching tool.

Cavities – or caries, as dentists call them – are infamously difficult to diagnose on X-rays. Cavities appear as slightly different shades of gray on a monochrome radiograph, and only 20%-40% are typically identified. It’s particularly easy to miss small cavities, which can often be reversed with treatment instead of needing to be filled more invasively.

Trained and validated by hundreds of dentists and honed with machine learning, Overjet can detect caries, hardened plaque and many other structures essential to oral health. The caries-identification module that UF students trained with highlights cavities in bright colors and aids in creating a treatment plan. The machine identification helped students – some of whom had seen few X-rays in their training so far – start to identify cavities more confidently on their own.

Preparing for AI in dentistry

In November of 2021, Mina Ghorbanifarajzadeh, D.M.D., a 2019 graduate of UF’s College of Dentistry and an early employee at Overjet, gave a presentation to her alma mater on the promise of the AI diagnostic tools Overjet was developing. 

“I thought it was important to make sure that the students had access to information before they graduated about how these tools were transforming their field,” said Ghorbanifarajzadeh, now senior clinical manager at Overjet. 

Ghorbanifarajzadeh’s visit eventually snowballed into an official collaboration between the College of Dentistry and Overjet. It was a natural fit. UF was pushing to expand AI education and research across the university, and Gohel was working with Isabel Garcia, D.D.S., M.P.H., dean of the College of Dentistry, to evaluate the machine learning tools on the market. Meanwhile, the college’s dean emeritus, Terri Dolan, D.D.S., had been named the chief dental officer at Overjet. She reached out to Garcia and offered to partner up.

In 2022, Overjet and the college partnered to incorporate the company’s AI education program, Overjet for Educators, into the UF curriculum.

“Part of what excites me about the tool is we have the opportunity to use it in the context of student training,” Garcia said. “Students can examine both the radiograph and the AI tool in the classroom, which provides an opportunity for the dental students to self-assess, which speeds up learning.”

Several students who have gone through the training have told Gohel they want to move into AI-focused careers. One student even decided to specialize in radiology because of their excitement about these dental AI systems.

“UF is the first dental school to incorporate our system into their curriculum,” Dolan said. “Students will graduate with a working knowledge of AI and computer vision.”

Patient upgrade

Soon, the College of Dentistry will adopt Overjet for patient care as well, eventually opening it up to the 100,000-plus patient visits the college provides every year. That means more accurate, earlier detection of cavities and better treatment plans for patients.

As part of the collaboration, Gohel tested the Overjet caries-identification module on over 1,000 cavities identified by expert dentists. In UF’s study, the module detected 91% of large caries and 69% of early-stage ones, making it far more sensitive than manual X-ray readings.

“I’m looking forward to incorporating Overjet technology into our patient care workflow. Students will benefit from visual and interactive diagnosis, and it will also have a positive impact on patient care,” Gohel said.

But the human element remains. Dentists must still perform clinical exams to verify the findings of the AI. Gohel warns her students against adopting automation bias, the tendency to trust machine systems too much. Just like human dentists, no dental AI system is perfect, and false positives and false negatives are both possible. 

Gohel prefers the term “augmented intelligence,” to emphasize how the program helps dentists; it doesn’t replace them. Overjet agrees.

“We really think our product assists clinicians in making clinical diagnoses by improving accuracy and consistency,” Dolan said.  

Now two years into the collaboration, the College of Dentistry is realizing the potential of Overjet’s systems to help them meet their three overlapping missions.

“It has the educational component, the research component, and the patient care component. We’re excited about all three elements,” Garcia said.

Eric Hamilton June 26, 2024