人工智能识别儿童及年轻成人上肢X线骨折
本研究开发并透明共享一个能检测儿童上肢多部位骨折的人工智能模型,基于58,846张上肢X线片训练和测试。强监督目标检测模型整体测试AUC为0.96、准确率89.7%、敏感度90.8%、特异度88.7%,表现优于弱监督模型;在夜间初步判读的病例上,AI准确率高于住院医师(89.4% vs. 85.1%,P=0.01)。结论认为目标检测AI模型识别儿童上肢骨折具有高准确率。
要点
- 强监督目标检测AI模型整体测试AUC 0.96、准确率89.7%、敏感度90.8%、特异度88.7%。
- 强监督模型AUC 0.96优于弱监督模型AUC 0.93(P<0.0001)。
- 夜间初步判读病例中AI准确率89.4%高于住院医师85.1%(P=0.01)。
核心结论
- 目标检测AI模型识别儿童上肢骨折具有高准确率。
研究设计与样本
- 研究纳入58,846张上肢X线片(手指/手、腕/前臂、肘、肱骨、肩/锁骨),来自14,873例儿童及年轻成人患者。
- 数据分为训练集(12,232例患者)、调优集(1,307例)、内部测试集(819例)和外部测试集(515例)。
- 骨折判定依据所有测试X线片的人工审阅,以及训练/调优集中报告经规则化自然语言处理(NLP)分类为骨折阳性的子集。
- 训练目标检测模型(Faster R-CNN,强监督)和图像分类模型(EfficientNetV2-Small,弱监督)检测骨折,并在测试数据上评估。
诊断标准与关键数值
- 强监督目标检测模型整体测试AUC 0.96(95%CI 0.95-0.97)、准确率89.7%(95%CI 88.0-91.3%)、敏感度90.8%(95%CI 88.5-93.1%)、特异度88.7%(95%CI 86.4-91.0%)。
- 弱监督模型AUC 0.93(95%CI 0.92-0.94),强监督模型AUC 0.96(95%CI 0.95-0.97)优于弱监督模型(P<0.0001)。
- 夜间初步判读病例中,AI准确率89.4%(95%CI 87.3-91.5%)高于住院医师准确率85.1%(95%CI 82.7-87.5%),P=0.01。
临床应用与阅片要点
- 作者结论认为目标检测AI模型识别儿童上肢骨折具有高准确率。
局限性
- 原文未报告局限性。
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
BACKGROUND: Pediatric fractures are challenging to identify given the different response of the pediatric skeleton to injury compared to adults, and most artificial intelligence (AI) fracture detection work has focused on adults.
OBJECTIVE: Develop and transparently share an AI model capable of detecting a range of pediatric upper extremity fractures.
MATERIALS AND METHODS: In total, 58,846 upper extremity radiographs (finger/hand, wrist/forearm, elbow, humerus, shoulder/clavicle) from 14,873 pediatric and young adult patients were divided into train (n = 12,232 patients), tune (n = 1,307), internal test (n = 819), and external test (n = 515) splits. Fracture was determined by manual inspection of all test radiographs and the subset of train/tune radiographs whose reports were classified fracture-positive by a rule-based natural language processing (NLP) algorithm. We trained an object detection model (Faster Region-based Convolutional Neural Network [R-CNN]; "strongly-supervised") and an image classification model (EfficientNetV2-Small; "weakly-supervised") to detect fractures using train/tune data and evaluate on test data. AI fracture detection accuracy was compared with accuracy of on-call residents on cases they preliminarily interpreted overnight.
RESULTS: A strongly-supervised fracture detection AI model achieved overall test area under the receiver operating characteristic curve (AUC) of 0.96 (95% CI 0.95-0.97), accuracy 89.7% (95% CI 88.0-91.3%), sensitivity 90.8% (95% CI 88.5-93.1%), and specificity 88.7% (95% CI 86.4-91.0%), and outperformed a weakly-supervised model (AUC 0.93, 95% CI 0.92-0.94, P < 0.0001). AI accuracy on cases preliminary interpreted overnight was higher than resident accuracy (AI 89.4% vs. 85.1%, 95% CI 87.3-91.5% vs. 82.7-87.5%, P = 0.01).
CONCLUSION: An object detection AI model identified pediatric upper extremity fractures with high accuracy.
引用
- Zech JR, Jaramillo D, Altosaar J, et al. Artificial intelligence to identify fractures on pediatric and young adult upper extremity radiographs. Pediatr Radiol. 2023. PMID: 37740031. doi:10.1007/s00247-023-05754-y
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。