腹部放射学中Gemini指南一致性决策支持评估
本研究基于4个腹部领域200个合成病例,评估Gemini 1.5 Flash的临床有用性、潜在危害、指南一致性及分诊表现。结果显示总体印象一致性为85.0%,最终分类一致性为77.5%,有用性中位数5分,危害中位数0分。但随病例难度增加,有用性下降、危害上升、一致性及分诊表现变差,支持其作为辅助决策工具而非复杂场景的自主管理。
要点
- Gemini在腹部病例中有用性评分高、危害评分低,可用于辅助决策。
- 最终分类一致性77.5%,复杂病例分诊适宜率降至66.2%,需人工复核。
- 报告可参考Gemini印象及建议,但应警惕高难度病例的潜在错误。
核心结论
- Gemini在腹部放射学中表现出高有用性、低总体危害和较强的指南一致性。
- 随着病例难度增加,性能下降,支持其作为辅助决策支持而非复杂或高风险场景的自主管理。
研究设计与样本
- 专家横断面研究,创建200个基于指南的合成腹部放射学病例,涵盖肾上腺、肾脏、肝脏和胰腺病变各50例。
- 病例难度前瞻性分为1、2、3级后,使用文本病例描述由Gemini 1.5 Flash生成印象、建议、理由和最终分类。
- 5名腹部放射科医生独立评价有用性(1-5分)和危害(0-3分),另2名独立评估印象一致性、最终分类一致性和分诊适宜性。
诊断标准与关键数值
- 总体印象一致性为85.0%(170/200),总体最终分类一致性为77.5%(155/200)。
- 有用性中位数为5(IQR=1),危害中位数为0(IQR=0)。
- 难度增加与有用性降低(p<0.001)、危害升高(p<0.001)、一致性降低(均p<0.001)及分诊变差相关。
- 适宜分诊率从1级病例的85.7%降至3级病例的66.2%,分诊分布随难度不同存在差异(p=0.012)。
临床应用与阅片要点
- 在日常阅片中可将Gemini输出作为腹部病变管理决策的辅助参考,尤其适用于低难度病例。
- 对于复杂或高风险病例,不能依赖Gemini的自主判断,应结合专家复核和指南确认。
- 关注Gemini最终分类与印象的建议,但在最终报告中保持放射科医生的独立判断。
局限性
- 研究采用合成病例而非真实临床病例,可能无法完全反映真实世界的复杂性。
- 仅使用文本病例描述,未利用实际影像图像,无法评估基于影像的决策支持能力。
- 仅评估Gemini 1.5 Flash单一模型,涉及单一中心专家评估,可能影响普遍适用性。
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: Large language models may help radiologists apply increasingly complex imaging guidelines, but their safety and clinical usefulness as decision-support tools in abdominal radiology remain uncertain. We evaluated the usefulness, potential harm, guideline concordance, and triage performance of Gemini across different levels of case difficulty.
MATERIALS AND METHODS: In this expert-based cross-sectional study, 200 synthetic, guideline-based abdominal radiology cases were created by consensus across four domains: adrenal, renal, liver, and pancreatic lesions (50 cases each). Case difficulty was prospectively assigned as level 1, 2, or 3 before LLM querying. Using text-only case descriptions, Gemini 1.5 Flash generated an impression, recommendation, rationale, and final category for each case. Five abdominal radiologists independently rated usefulness (1-5) and harm (0-3). Two abdominal radiologists independently assessed impression concordance, final-category concordance, and triage appropriateness relative to the reference standard.
RESULTS: Median usefulness was 5 (IQR, 1), and median harm was 0 (IQR, 0). Overall impression concordance was 85.0 % (170/200), and final-category concordance was 77.5 % (155/200). Greater case difficulty was associated with lower usefulness (p < 0.001), higher harm (p < 0.001), lower concordance (both p < 0.001), and worse triage performance. Triage distributions differed by case difficulty (p = 0.012), with appropriate triage declining from 85.7 % in level 1 cases to 66.2 % in level 3 cases.
CONCLUSION: Gemini showed high perceived usefulness, low overall harm, and generally strong guideline concordance in abdominal radiology. However, declining performance with greater case difficulty supports its use as assistive decision support rather than autonomous management in complex or high-risk scenarios.
引用
- Kaya, Sabet. Guideline-concordant decision support in abdominal radiology: An expert-based evaluation of Gemini. Eur J Radiol. 2026. PMID: 42607501. doi:10.1016/j.ejrad.2026.113166
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。