Machine learning techniques for periodontitis and dental caries detection: A narrative review

Area of Science:

• Dental diagnostics
• Machine learning applications
• Artificial intelligence in healthcare

Background:

• Periodontitis and dental caries are prevalent oral health issues requiring early diagnosis to prevent tooth loss.
• Current diagnostic methods rely on visual inspection, probing, and radiographs by experienced dentists.
• The clinical suitability of machine learning (ML) algorithms for automated dental diagnostics remains under investigation.

Purpose of the Study:

• To review and analyze machine learning applications for automated periodontitis and dental caries detection.
• To identify research challenges and limitations in current ML methods for dental diagnostics.
• To explore pathways for developing robust ML systems for clinical use and point-of-care testing.

Main Methods:

• Extensive literature search from 2015-2022 in PubMed, IEEE Xplore, and ScienceDirect.
• Inclusion of 55 studies focusing on ML for detecting periodontitis, dental caries, and related issues.
• Analysis of ML techniques and their diagnostic accuracy based on selected studies.

Main Results:

• Most ML studies utilize radiograph images for dental diagnostics.
• A subset of studies demonstrates high diagnostic accuracy using smartphone-captured images.
• ML applications cover periodontitis, dental caries, apical lesions, periodontal bone loss, and vertical root fractures.

Conclusions:

• Smartphone-based ML models can achieve good diagnostic accuracy for dental issues.
• Utilizing smartphone images can reduce clinical diagnosis costs and enhance user interaction.
• Further research is needed to overcome limitations and establish robust ML systems for widespread clinical adoption.