Artificial intelligence for dental caries diagnosis: translating algorithms to clinical practice

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Dental caries is the most common chronic disease globally, causing significant healthcare costs.
• Current diagnostic methods lack sensitivity and have high variability, especially for early-stage lesions.
• Artificial intelligence (AI) offers potential but faces a translational gap for clinical use.

Purpose of the Study:

• To review the AI development pipeline for caries diagnosis, from data to deployment.
• To identify bottlenecks hindering AI clinical translation.
• To propose a roadmap for developing reliable AI decision-support systems.

Main Methods:

• Synthesis of the AI development pipeline for caries diagnosis.
• Identification of challenges: domain shift, annotation inconsistency, limited multimodal data, and varied standards.
• Review of emerging strategies: multi-center data, self-supervised learning, domain adaptation.

Main Results:

• Significant challenges persist in translating AI for dental caries diagnosis into clinical practice.
• Existing AI models often struggle with domain shift and inconsistent data annotation.
• Emerging strategies show promise but are underutilized.

Conclusions:

• A paradigm shift towards quantitative, risk-based staging is needed, aligning with minimally invasive dentistry.
• Standardized data, rigorous validation, and human-centered integration are crucial for trustworthy AI.
• The proposed roadmap aims to reduce the global burden of untreated caries through effective AI deployment.