多模态大语言模型用于儿童骨龄评估:一项比较准确性分析
研究以180张左手腕部X线片(0-18岁,RSNA儿童骨龄数据集)比较三种多模态大语言模型与放射科医师共识调整骨龄的准确性,Gemini 2.5 Pro误差最低(MAE 28.48个月),但所有模型95%一致性界限均超过±4.5年,作者认为当前误差不适合临床或法医学决策。
要点
- Gemini 2.5 Pro的MAE为28.48个月,低于ChatGPT-4.5的35.38个月和ChatGPT-o4-mini-high的35.87个月(均P<.001)。
- 所有模型95%一致性界限均超过±4.5年,误差水平不适合临床或法医学决策。
- 多模态大语言模型无需任务特定训练即可从X线片估算骨龄,但当前2-3年误差仍不理想。
核心结论
- 多模态大语言模型无需任务特定训练即可直接从X线片估算骨龄,但当前2-3年的误差和宽泛的误差范围仍不适合临床或法医学决策;在替代专用骨龄评估软件或专家阅片前,需要视觉编码器能力和领域特定微调的进步。
研究设计与样本
- 比较准确性分析,使用180张左手腕部X线片(0-18岁,RSNA儿童骨龄数据集),由ChatGPT-o4-mini-high、ChatGPT-4.5和Google Gemini 2.5 Pro三种模型使用相同提示词分析,参考标准为放射科医师共识调整骨龄。
诊断标准与关键数值
- 准确性指标为平均绝对误差(MAE)和均方根误差(RMSE):Gemini 2.5 Pro的MAE 28.48个月、RMSE 35.23个月;ChatGPT-4.5的MAE 35.38个月、RMSE 43.64个月;ChatGPT-o4-mini-high的MAE 35.87个月、RMSE 45.43个月(Gemini与后两者比较均P<.001)。
临床应用与阅片要点
- 作者认为当前多模态大语言模型误差为2-3年且误差范围宽,不适合临床或法医学决策,在替代专用骨龄评估软件或专家阅片前需要视觉编码器能力和领域特定微调的进步。
局限性
- 所有模型95%一致性界限均超过±4.5年。
- 作者指出当前2-3年误差仍不适合临床或法医学决策。
- 作者认为在替代专用骨龄评估软件或专家阅片前,需要视觉编码器能力和领域特定微调的进步。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
RATIONALE AND OBJECTIVES: Bone age assessment guides endocrine and orthopedic care in children, yet manual Greulich‑Pyle reading is subjective, and commercial deep‑learning tools are costly. We examined whether readily accessible closed‑source multimodal large language models can provide clinically acceptable bone age assessment accuracy.
MATERIALS AND METHODS: A balanced set of 180 left-hand wrist radiographs (ages 0-18 years; RSNA Pediatric Bone Age dataset) was analyzed by three models (ChatGPT‑o4-mini‑high, ChatGPT‑4.5, and Google Gemini 2.5 Pro) using an identical prompt. Reference ages were adjusted bone ages by consensus of radiologists. Accuracy metrics were mean absolute error (MAE) and root‑mean‑square error (RMSE); agreement was assessed with Bland-Altman analysis. Model performance was compared with a Friedman test followed by Bonferroni‑adjusted Wilcoxon signed‑rank tests.
RESULTS: Gemini 2.5 Pro yielded the lowest error (MAE 28.48 months, RMSE 35.23 months), comparing with ChatGPT‑4.5 (MAE 35.38 months, RMSE 43.64 months) and ChatGPT‑o4-mini‑high (MAE 35.87 months, RMSE 45.43 months; both P < .001). Bias was -9.38 months for Gemini, -4.80 months for ChatGPT‑4.5, and +5.26 months for ChatGPT‑o4-mini‑high; 95% limits of agreement exceeded ±4.5 years for every model.
CONCLUSION: Multimodal Large Language Models can estimate bone age directly from radiographs without task‑specific training, but current errors of 2-3years and wide errors remain unsuitable for clinical or medico‑legal decision‑making. Advances in vision encoder capacity and domain‑specific fine‑tuning are needed before LLMs can replace dedicated BAA software or expert readers.
引用
- Büyüktoka RE, Salbas A. Multimodal Large Language Models for Pediatric Bone-Age Assessment: A Comparative Accuracy Analysis. Acad Radiol. 2025. PMID: 40803966. doi:10.1016/j.acra.2025.07.058
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。