GPT-5-Thinking 在眶部及头颈肿瘤 MRI 报告中的双中心评估
本研究为双中心回顾性研究,纳入 1000 例经病理证实的眶部或头颈肿瘤患者,以治疗前 MRI 报告为输入,评估 GPT-5-Thinking 的良恶性分类与前三位鉴别诊断效能。结果显示模型诊断准确率与亚专科医师相仿,并可提升普放医师独立诊断的准确率。
要点
- GPT-5-Thinking 对眶部肿瘤的诊断准确率与亚专科医师相当。
- 普放医师在使用 GPT-5-Thinking 辅助后,头颈肿瘤诊断准确率由 47.0% 升至 61.0%。
- 该模型基于叙述性 MRI 报告文本输出诊断,不直接阅读影像图像。
核心结论
- GPT-5-Thinking 基于 MRI 报告对眶部及头颈肿瘤的诊断准确率与亚专科医师相似。
- GPT-5-Thinking 辅助与普放医师诊断准确率提高相关。
研究设计与样本
- 双中心回顾性研究,包括一个普放实践中心和一个亚专科实践中心。
- 纳入 1000 例经病理证实的眶部或头颈肿瘤患者,均有治疗前 MRI 报告。
- Phase 1 比较模型与常规临床报告诊断的准确率。
- Phase 2 由 7 名普放医师在无辅助和有 GPT-5-Thinking 辅助下解读指定肿瘤报告。
- 使用 McNemar 检验和混合效应 logistic 回归模型进行统计分析。
关键影像表现/征象
- GPT-5-Thinking 对眶部肿瘤前一位诊断准确率为 76.7%,亚专科医师为 78.3%,差异无统计学意义。
- GPT-5-Thinking 对头颈肿瘤前一位诊断准确率为 67.5%,亚专科医师为 68.0%,差异无统计学意义。
- GPT-5-Thinking 对眶部肿瘤前一位诊断准确率高于普放医师常规报告,为 74.0% 对 68.7%。
- GPT-5-Thinking 对头颈肿瘤前一位诊断准确率高于普放医师常规报告,为 64.0% 对 27.0%。
诊断标准与关键数值
- 眶部肿瘤:GPT-5-Thinking 前一位诊断准确率 76.7%,亚专科医师 78.3%,P = .64。
- 头颈肿瘤:GPT-5-Thinking 前一位诊断准确率 67.5%,亚专科医师 68.0%,P = .92。
- 眶部肿瘤:普放医师无辅助时前一位准确率 61.4%,有辅助时升至 70.3%,P < .001。
- 头颈肿瘤:普放医师无辅助时前一位准确率 47.0%,有辅助时升至 61.0%,P < .001。
- 样本平均年龄 51 岁 ± 17.0,其中男性 520 例。
临床应用与阅片要点
- GPT-5-Thinking 可用于辅助普放医师解读眶部及头颈肿瘤 MRI 报告,缩小与亚专科医师之间的诊断差距。
- 模型输入为叙述性 MRI 报告文本而非影像图像,适用于已有结构化或叙述性报告的场景。
局限性
- 双中心回顾性设计,样本为便利系列,可能存在选择偏倚。
- 未直接评估模型对影像图像本身的解读能力,仅基于报告文本。
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
Background Identifying tumor histologic subtypes in complex regions is challenging for generalist radiologists. Whether reasoning large language models can bridge this expertise gap by interpreting radiologic descriptions remains underexplored. Purpose To evaluate whether Generative Pretrained Transformer (GPT)-5-Thinking can perform similarly to subspecialists and help bridge the generalist expertise gap in interpretation of MRI scans of orbital and head-and-neck tumors. Materials and Methods This dual-center retrospective study was performed at a generalist practice center and a subspecialist practice center and included convenience series of patients with pathologically confirmed orbital or head-and-neck tumors who had pretreatment MRI reports. GPT-5-Thinking processed narrative MRI reports to output benign-malignant classification and top-3 differential diagnoses. Phase 1 compared model accuracy with routine clinical report diagnoses in generalist and subspecialist practice settings. In Phase 2, seven generalists interpreted selected tumor reports twice, without and with GPT-5-Thinking assistance. McNemar tests and mixed-effects logistic regression models were used for analysis. Results This study included 1000 patients (mean age ± SD, 51 years ± 17.0; 520 men). In phase 1, the top-1 accuracy of GPT-5-Thinking was similar to that of subspecialists in patients with orbital tumors (76.7% [230 of 300] vs 78.3% [235 of 300]; P = .64) and head-and-neck tumors (67.5% [135 of 200] vs 68.0% [136 of 200]; P = .92). GPT-5-Thinking had higher top-1 accuracy than did routine generalists for patients with orbital tumors (74.0% [222 of 300] vs 68.7% [206 of 300]; P = .03) and head-and-neck tumors (64.0% [128 of 200] vs 27.0% [54 of 200]; P < .001). In phase 2, GPT-5-Thinking assistance was associated with an increase in generalist top-1 diagnostic accuracy from 61.4% (430 of 700) to 70.3% (492 of 700) (P < .001) in patients with orbital tumors and from 47.0% (329 of 700) to 61.0% (427 of 700) (P < .001) in patients with head-and-neck tumors. Conclusion Using MRI reports of patients with orbital or head-and-neck tumors, Generative Pretrained Transformer (GPT)-5-Thinking showed diagnostic accuracy similar to that of subspecialists; GPT-5-Thinking assistance was also associated with higher generalist diagnostic accuracy. © RSNA, 2026 Supplemental material is available for this article.
引用
- Li, Xiao, Qu, et al. Bridging the Generalist-Subspecialist Gap with GPT-5-Thinking: Dual-Center Evaluation in Orbital and Head-and-Neck Tumor MRI Reports. Radiology. 2026. PMID: 42742382. doi:10.1148/radiol.253579
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。