A benchmark multimodal oro-dental dataset for large vision-language models

2025-11-07

AI Summary

Current oral AI research is hindered by the scarcity of large-scale, multimodal datasets that reflect clinical complexity. To address this, we introduce OralMed-DB—the first standardized, large-model-oriented dental multimodal benchmark—comprising 4,800 patients and 8,775 clinical examinations, with aligned intraoral images, radiographs (e.g., panoramic and periapical X-rays), and structured electronic health record text. The benchmark supports two clinically relevant tasks: classification of six common dental abnormalities and generation of diagnostic reports. We fine-tune vision-language models—including Qwen-VL (3B and 7B variants)—integrating radiographic analysis and natural language understanding capabilities. Experimental results demonstrate statistically significant improvements over strong baselines and GPT-4o on both tasks, validating OralMed-DB’s efficacy, generalizability, and clinical applicability. This work establishes a foundational infrastructure for advancing practical, deployable AI in dentistry.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Multi-modal Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

The advancement of artificial intelligence in oral healthcare relies on the availability of large-scale multimodal datasets that capture the complexity of clinical practice. In this paper, we present a comprehensive multimodal dataset, comprising 8775 dental checkups from 4800 patients collected over eight years (2018-2025), with patients ranging from 10 to 90 years of age. The dataset includes 50000 intraoral images, 8056 radiographs, and detailed textual records, including diagnoses, treatment plans, and follow-up notes. The data were collected under standard ethical guidelines and annotated for benchmarking. To demonstrate its utility, we fine-tuned state-of-the-art large vision-language models, Qwen-VL 3B and 7B, and evaluated them on two tasks: classification of six oro-dental anomalies and generation of complete diagnostic reports from multimodal inputs. We compared the fine-tuned models with their base counterparts and GPT-4o. The fine-tuned models achieved substantial gains over these baselines, validating the dataset and underscoring its effectiveness in advancing AI-driven oro-dental healthcare solutions. The dataset is publicly available, providing an essential resource for future research in AI dentistry.

Problem

• Addressing the lack of large multimodal datasets for oral healthcare AI
• Classifying six oro-dental anomalies using vision-language models
• Generating complete diagnostic reports from multimodal dental inputs

Innovation

• Large multimodal dataset for dental AI
• Fine-tuned vision-language models on dental data
• Benchmarking models for dental anomaly classification