Deep Learning Applications in Dental Image-Based Diagnostics: A Systematic Review

Area of Science:

• Dental Informatics
• Artificial Intelligence in Medicine
• Systematic Review and Meta-analysis

Background:

• Artificial intelligence (AI) is increasingly utilized in dentistry for diagnosis, treatment planning, and prognosis.
• This review systematically assesses AI models in dentistry, evaluating their performance, limitations, and future integration potential.

Purpose of the Study:

• To identify and evaluate AI models applied in dentistry.
• To assess the diagnostic accuracy and predictive performance of these AI models.
• To discuss the challenges and ethical considerations for AI adoption in dental practice.

Main Methods:

• Systematic literature search of PubMed, Scopus, and Cochrane Library.
• Meta-analysis of 20 selected studies out of 947 identified papers.
• Assessment of diagnostic accuracy, predictive performance, and potential biases of AI models.

Main Results:

• AI models achieved an average diagnostic accuracy of 82%, predominantly using artificial neural networks (ANNs) and convolutional neural networks (CNNs).
• Significant improvements in diagnosing dental caries compared to traditional methods were observed.
• AI shows promise in detecting various conditions like bone loss, lesions, cysts, and in orthodontic assessments, but faces challenges in data bias, cost, and data security.

Conclusions:

• AI holds transformative potential for dentistry, enhancing diagnostic precision and treatment planning.
• Critical evaluation of AI's benefits, drawbacks, and ethical implications is necessary before widespread clinical adoption.
• Future research should address barriers related to data, cost, and security to facilitate effective AI utilization in dental healthcare.