Artificial intelligence outperforms experienced clinicians in radiograph diagnostics study

October 7, 2020

A study published in early September finds that artificial intelligence-powered diagnostic platforms can identify tooth decay from x-rays more accurately and consistently than human clinicians.

Dental artificial intelligence company Pearl recently completed a study comparing the diagnostic accuracy and consistency between its Second Opinion artificial intelligence (AI)-powered software platform and experienced clinicians with the software platform coming out on top.

The study had the software and three dentists evaluate the same set of 8767 bitewing and periapical x-rays. Results showed the human dentists agreed 79% of the time when there was no decay, but only agreed upon the presence of decay in 4.2% of the evaluated radiographs. The study found it was common for two of the three dentists to identify decay in the same x-ray, but often the other dentist did not spot the same decay.

The AI system was able to more accurately identify the decay, with its ability to evaluate radiographs based on the study of millions of x-rays that had been previously reviewed by human doctors. While the results are encouraging for the efficacy of AI in dentistry, the researchers point to a need for additional study to determine if the AI system is better at identifying specific types of caries better than others.

The full study, titled "Can a Computer Identify Carious Lesions in Dental X-Rays As Accurately
As Humans? An exploratory study comparing diagnostic assessments performed by humans and a specialized computer vision system" can be found here.