A study published by ScienceDirect in September validated an earlier AI model to detect molar incisor hypomineralization (MIH), a global oral health issue.Last year, a groundbreaking study introduced an open-access AI model for detecting MIH, which affects about 14 per cent of the global population, using digital photographs.
Researchers at the Department of Conservative Dentistry and Periodontology at LMU Munich have externally validated the model, achieving up to 94.3 per cent accuracy for image-based detection of EH/MIH, further enhancing the method’s efficacy.
The researchers behind the new study acknowledged its “strengths” and “limitations.”
“As a strength, it should be emphasized that for the first time, an AI-based model for automated detection of EH/MIH in dental photographs at the pixel level was subjected to external validation,” the study stated.
However, limitations include image quality and assessments based on subjective judgments, categorizing images as “acceptable” or “good.” Only image size was objectively recorded from the metadata.
Additionally, there is no universally accepted standard for image quality in studies like this.
In the earlier study, the AI-based algorithm analyzed an image set of 18,179 anonymous photographs, resulting in 34,710 pathological findings.
