Related Experiment Video
Updated: Aug 5, 2026
07:32
Measuring Maxillary Posterior Tooth Movement: A Model Assessment using Palatal and Dental Superimposition
Published on: February 23, 2024
Performance of multimodal large language models in interpreting lateral cephalometric superimpositions: A comparative
Viet Anh Nguyen1, Thi Minh Anh Ha2, Thi Thu Huong Nguyen1
1School of Dentistry, Hanoi Medical University, Hanoi, Viet Nam.
International Orthodontics
|July 27, 2026
Summary
Multimodal large language models (LLMs) show limited performance in interpreting orthodontic cephalometric superimpositions compared to human experts. These AI models should not replace professional orthodontic assessment without expert oversight.
Area of Science:
- Artificial Intelligence in Dentistry
- Medical Image Analysis
- Orthodontic Diagnostics
Background:
- Multimodal large language models (LLMs) offer potential for free-text interpretation of clinical images.
- The efficacy of LLMs in orthodontic cephalometric superimposition analysis remains unexplored.
Purpose of the Study:
- To evaluate the performance of three leading LLMs in interpreting orthodontic cephalometric superimpositions.
- To compare LLM interpretations against those of an orthodontic resident and senior orthodontists.
Main Methods:
- Analysis of 90 lateral cephalometric superimposition images (nongrowing, growing, orthognathic cases).
- Zero-shot interpretation by ChatGPT, Gemini, Claude, and a second-year orthodontic resident using identical prompts.
- Scoring by senior orthodontists using a 16-item rubric against adjudicated reference interpretations.
Main Results:
- Significant differences in total scores among all methods (Friedman P<0.001).
- Resident achieved a median score of 30.5, significantly outperforming ChatGPT (17), Gemini (12), and Claude (9.5).
- All LLMs performed significantly below the resident across all domains and case types (adjusted P<0.001).
Conclusions:
- Tested multimodal LLMs demonstrated substantially lower performance than an orthodontic resident in zero-shot cephalometric superimposition interpretation.
- Current LLMs require expert review and should not be utilized as standalone diagnostic tools in this context.
