LLMs for Jaw Lesion Diagnosis on Panoramic Radiographs: Gemini, ChatGPT, Copilot Compared

Gemini achieved highest mean total scores for radiolucent (11.49) and mixed (9.01) jaw lesions on panoramic radiographs, ChatGPT for radiopaque (10.93), while Copilot scored lowest; significant differences across LLMs (p<0.05).

• Gemini scored highest for radiolucent (mean 11.49) and mixed (9.01) lesions, ChatGPT for radiopaque lesions (10.93).
• Copilot showed the lowest total scores across all lesion types.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.