DentalGPT: Incentivizing Multimodal Complex Reasoning in Dentistry
Zhenyang Cai, Jiaming Zhang, Junjie Zhao, Ziyi Zeng, Yanchao Li, Jingyi Liang, Junying Chen, Yunjin Yang, Jiajun You, Shuzhi Deng, Tongfei Wang, Wanting Chen, Chunxiu Hao, Ruiqi Xie, Zhenwei Wen, Xiangyi Feng, Zou Ting, Jin Zou Lin, Jianquan Li, Guangjun Yu, Liangyi Chen, Junwen Wang
2025-12-15
Summary
This paper introduces DentalGPT, a new artificial intelligence model designed to understand and interpret dental images and information, ultimately aiming to help with automated dental healthcare.
What's the problem?
Current AI models, called multimodal large language models, aren't very good at recognizing the small, important details in dental images, and they struggle with the complex reasoning needed to make accurate diagnoses. They lack the specific knowledge needed for dentistry, hindering their ability to be truly helpful in a clinical setting.
What's the solution?
The researchers created DentalGPT by first building a huge collection of over 120,000 dental images, each with detailed descriptions pointing out key visual features. They then trained an AI model on this data to improve its ability to 'see' and understand dental conditions. Finally, they used a technique called reinforcement learning to further enhance the model’s reasoning skills, allowing it to answer complex questions about dental images. The final model is relatively small, with only 7 billion parameters.
Why it matters?
This work is important because it shows that by focusing on a specific medical field, like dentistry, and providing high-quality training data, we can create AI models that perform much better than general-purpose models. This could lead to tools that assist dentists with diagnosis, treatment planning, and ultimately, improve patient care.
Abstract
Reliable interpretation of multimodal data in dentistry is essential for automated oral healthcare, yet current multimodal large language models (MLLMs) struggle to capture fine-grained dental visual details and lack sufficient reasoning ability for precise diagnosis. To address these limitations, we present DentalGPT, a specialized dental MLLM developed through high-quality domain knowledge injection and reinforcement learning. Specifically, the largest annotated multimodal dataset for dentistry to date was constructed by aggregating over 120k dental images paired with detailed descriptions that highlight diagnostically relevant visual features, making it the multimodal dataset with the most extensive collection of dental images to date. Training on this dataset significantly enhances the MLLM's visual understanding of dental conditions, while the subsequent reinforcement learning stage further strengthens its capability for multimodal complex reasoning. Comprehensive evaluations on intraoral and panoramic benchmarks, along with dental subsets of medical VQA benchmarks, show that DentalGPT achieves superior performance in disease classification and dental VQA tasks, outperforming many state-of-the-art MLLMs despite having only 7B parameters. These results demonstrate that high-quality dental data combined with staged adaptation provides an effective pathway for building capable and domain-specialized dental MLLMs.