Dental News - Study highlights how artificial intelligence can be used for detection of caries

Search Dental Tribune

Study highlights how artificial intelligence can be used for detection of caries

A study from researchers at Charité—Universitätsmedizin Berlin has sought to measure the impact that artificial intelligence has on the ability of dentists to detect caries. (Image: Nakleyka/Shutterstock)
Brendan Day, Dental Tribune International

Brendan Day, Dental Tribune International

Wed. 3. November 2021

save

BERLIN, Germany: Though artificial intelligence (AI) is being increasingly integrated into a variety of dental products and services, the body of literature evaluating its perceived benefits is scarce. To help rectify this, researchers from Charité—Universitätsmedizin Berlin have recently published the results of a randomised controlled trial they conducted. These results demonstrate that AI can increase the diagnostic accuracy of dentists.

As Dental Tribune International (DTI) has previously reported, AI technologies are steadily being adopted by dental practices aiming to digitise and streamline their workflows. From initial consultations, diagnosis and treatment planning through to surgical procedures and postoperative care, a range of dental tasks can now be augmented by the various AI solutions that have been developed in recent years.

The performance of these AI-powered tools in medical and dental settings, however, has rarely been tested in clinical trials. As a result, the real impact of AI on the decision-making and diagnostics processes of dental practitioners remains somewhat unknown. This lack of clarity can carry over into decisions regarding available courses of treatment and their advantages.

The research team thus commenced a trial using dentalXrai Pro, a software program that allows dental practitioners to analyse radiographs based on AI. The dentalXrai Pro project was co-founded at Charité by Prof. Falk Schwendicke, head of the Department of Oral Diagnostics, Digital Health and Health Services Research, and has since been spun off into a start-up simply titled dentalXrai.

The AI-utilising software was employed by the 22 participating dentists to support their detection of caries on 20 bitewing images randomly chosen from a pool of 140. Of the 20 images analysed by each dentist, ten were examined with the assistance of this AI tool, whereas the other ten were not.

“Gathering evidence to better evaluate the benefits that AI can deliver dentists is at the core of what we do” – Prof. Falk Schwendicke

According to the study authors, their hypothesis—that dentists using AI would be significantly more accurate than those not using AI—proved to be partially correct. In their discussion, they noted that “using AI significantly increased dentists’ sensitivity, especially on enamel caries lesions, but did not greatly alter specificity; on more advanced lesions AI did not impact on accuracy at all”. They stated that it was likely the AI was more helpful in situations where changes between images were miniscule, and that it played a lesser role when carious developments were significant and relatively easy to notice.

“Our results demonstrate that combining the AI model performance with human expertise can reach accuracies which are beyond those of the AI itself […] or the human experts on their own,” the authors wrote.

It was also noted, however, that using the AI software led to an increased likelihood of the dentists deciding to use invasive restorative therapy to treat the carious lesions. “In this sense, using an AI support to improve sensitivity may increase the risk of type I errors and overtreatment,” the authors remarked, adding that it could be beneficial for the dental industry to provide evidence-backed treatment recommendations for lesions of various depths. In their view, this would ideally lead to “better, not necessarily more invasive care”.

Prof. Schwendicke confirmed that further studies regarding the dentalXrai Pro software are already being planned.

“We are already planning on examining a different sample cohort using the updated version of this software that will be available early next year,” he told DTI. “Gathering evidence to better evaluate the benefits that AI can deliver dentists is at the core of what we do,” he explained.

The study, titled “Artificial intelligence for caries detection: Randomized trial”, was published online on 14 October 2021 in the Journal of Dentistry, ahead of inclusion in the December 2021 issue.

Tags:
To post a reply please login or register
advertisement
advertisement