AI-Based Detection of Dental Caries in Children

Background:

Dental caries is a prevalent oral health issue among children, affecting their quality of life and overall health. Early detection is critical for the effective management and prevention of caries progression. Traditional diagnostic methods, such as visual examinations and radiographs, often rely on the examiner's skill and may miss early-stage lesions or lead to misdiagnoses. To address these challenges, advances in artificial intelligence (AI), particularly deep learning and convolutional neural networks (CNNs), have shown promising results in medical imaging and diagnostics. This study aims to explore the use of AI in enhancing the early detection and categorization of dental caries in pediatric patients.

Study Objectives:

The primary objective of this study is to evaluate the diagnostic accuracy of an AI-based system for detecting and classifying dental caries in pediatric patients using intraoral digital imaging. The secondary objectives include:

Assessing the reproducibility and reliability of the AI model when compared to traditional clinical examinations.

Investigating the feasibility and potential benefits of integrating AI into routine pediatric dental practice.

Analyzing the impact of AI in reducing diagnostic variability among different examiners.

Study Hypotheses:

Primary Hypothesis: The AI-based model will demonstrate diagnostic accuracy comparable to or exceeding that of traditional clinical visual examination in detecting dental caries among pediatric patients.

Secondary Hypothesis: The AI model will exhibit high reproducibility and inter-examiner reliability, reducing the variability inherent in conventional diagnostic approaches.

Methodology:

Trial Design:

This is a prospective study designed to collect data before the index test (AI-based caries detection system) and reference standard (clinical examination) are applied. The study will compare the diagnostic performance of the AI system against conventional clinical methods, focusing on key metrics such as sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall diagnostic accuracy.

Participants:

Pediatric patients aged 6-9 years, presenting at the Pediatric Dentistry Department of Cairo University, will be recruited for this study. Participants will be selected based on their ability to cooperate during clinical examination and imaging procedures. Patients with visible dental caries will be included, while those with systemic conditions affecting oral health or those who cannot cooperate will be excluded.

Interventions:

AI-based System: Participants will undergo intraoral digital imaging, which will be analyzed using an AI model developed to detect and classify dental caries. The AI system employs deep learning techniques and CNNs to identify caries in early stages, including non-cavitated lesions.

Traditional Clinical Examination: A trained dentist will perform a standard visual and tactile clinical examination to assess dental caries, using a mirror and probe.

Data Collection:

Clinical data will include demographic details (age, gender, medical history) and dental examination findings. The AI system's diagnostic results will be compared with the clinical examination outcomes. The study will also assess the feasibility of integrating the AI model into routine practice by evaluating its ease of use and the time required for analysis.

Study Duration:

The study will be conducted over a six-month period, from January to June 2024. Participants will be monitored throughout the study for any adverse effects or concerns.

Informed Consent:

Before participating in the study, informed consent will be obtained from the parents or guardians of all pediatric patients. Participants will be fully informed about the procedures and their right to withdraw at any time without any consequences.

Safety and Monitoring:

While the AI system is not expected to pose any direct risk to the participants, the study will be closely monitored by healthcare providers to ensure patient safety. Participants' well-being will be regularly assessed during the course of the study, and any concerns raised by the participants or their guardians will be addressed promptly.

Expected Outcomes:

The AI-based system is expected to perform comparably to or better than traditional clinical methods in detecting dental caries, particularly in identifying early lesions that may otherwise go unnoticed.

The system's ability to standardize diagnosis and reduce inter-examiner variability will be evaluated.

The feasibility of incorporating AI-based detection into routine pediatric dental practice will be assessed, particularly its potential to improve workflow efficiency and reduce diagnostic time.

Statistical Analysis:

Data will be analyzed using standard statistical methods. Sensitivity, specificity, PPV, NPV, and the area under the curve (AUC) will be calculated to evaluate the diagnostic performance of the AI system compared to clinical examination. Reproducibility and inter-examiner reliability will be assessed using statistical tests for agreement, such as the kappa coefficient.

Conclusion:

This study aims to provide valuable insights into the effectiveness and practicality of AI-based dental caries detection in pediatric patients. By comparing the AI model with traditional diagnostic methods, we hope to demonstrate the potential of AI to enhance early caries detection, reduce diagnostic variability, and improve overall dental care for children.