LMCD-OR: a large-scale, multilevel categorized diagnostic dataset for oral radiography

Area of Science:

• Oral medicine and radiology
• Artificial intelligence in healthcare
• Medical imaging analysis

Background:

• Digital dentistry increasingly uses AI for improved clinical outcomes.
• A lack of comprehensive benchmark datasets hinders AI development in oral diagnostics.
• Oral radiograph analysis is crucial for dental disease diagnosis.

Purpose of the Study:

• To introduce LMCD-OR, a novel, large-scale dataset of oral radiograph images.
• To provide a benchmark dataset for training and validating AI models in oral diagnostics.
• To support advancements in AI-driven dental disease classification.

Main Methods:

• Collected 3,818 DICOM oral X-ray images from local medical institutions.
• Meticulously annotated images for primary dental services and secondary disease diagnoses.
• Utilized four state-of-the-art visual neural network models for dataset validation.

Main Results:

• The LMCD-OR dataset supports multiclassification models for oral disease diagnostics.
• Benchmark models demonstrated robust performance on the dataset.
• The dataset facilitates improved precision and scope in AI-driven oral diagnostics.

Conclusions:

• LMCD-OR addresses the need for comprehensive benchmark datasets in AI-driven dentistry.
• The dataset is valuable for training and validating AI models for oral disease diagnosis.
• LMCD-OR is publicly available to foster research and development in digital dentistry.