| 俞超丽,蒋迪华,吴梅.影像组学辅助X线诊断胫骨平台骨折效能研究[J].浙江中西医结合杂志,2025,35(11): |
| 影像组学辅助X线诊断胫骨平台骨折效能研究 |
| Research on the Efficacy of Radiomics-Assisted X-ray Diagnosis for Tibial Plateau Fractures YU Chaoli, JIANG Dihua, WU Mei Department of Radiology, Fuyang District Traditional Chinese Medicine Orthopaedic Hospital of Hangzhou, Hangzhou, Zhejiang, 311400, China. |
| 投稿时间：2025-05-22 修订日期：2025-08-21 |
| DOI： |
| 中文关键词: 人工智能 影像组学 X射线 胫骨平台骨折 |
| 英文关键词:Artificial intelligence Radiomics X-ray Tibial plateau fractures |
| 基金项目: |
|
| 摘要点击次数: 71 |
| 全文下载次数: 0 |
| 中文摘要: |
| 目的 探究人工智能影像组学在胫骨平台骨折X线诊断中的应用价值。方法 本研究纳入杭州市富阳中医骨伤医院收治的1580例膝关节外伤患者,并将其分为训练组1263例和验证组317例。通过标注X线影像中的骨折区域,构建人工智能影像组学诊断模型,并采用受试者工作特征曲线(Receiver Operation Characteristic, ROC)、ROC曲线下面积(Area Under the Curve, AUC)、校准曲线及决策曲线分析(Decision Curve Analysis, DCA)等多重指标进行模型效能验证。结果 训练组与验证组的AUC值分别达1.0000和0.9748,校准曲线证实模型预测值与实际诊断结果高度吻合,DCA分析进一步证实其临床净收益显著。诊断对比实验中,医师独立诊断组的准确率(0.9085)与召回率(0.8929)均低于模型独立诊断组(准确率0.9306,召回率0.9071),而模型辅助医师诊断组的诊断效能相对最高(准确率0.9590,召回率0.9429)。结论 影像组学模型能显著提升X线诊断胫骨平台骨折的准确率和效率,有效降低漏诊与误诊风险,为骨折的早期精准诊断提供了可靠的技术支持。 |
| 英文摘要: |
| Objective This research aimed to evaluate the application value of AI-based radiomics in the X-ray diagnosis of tibial plateau fractures. Methods A total of 1580 patients diagnosed at Fuyang District Traditional Chinese Medicine Orthopaedic Hospital of Hangzhou were included in this research and divided into a training group (n=1263) and a validation group (n=317). AI radiomics diagnostic model was constructed by segmenting fracture regions on X-ray images. Model performance was validated using multiple metrics including receiver operating characteristic curves (ROC), area under the ROC curve (AUC), calibration curves, and decision curve analysis (DCA). Results The AUC values for the training group and validation group reached 1.0000 and 0.9748 respectively. Calibration curves confirmed a high concordance between model predictions and actual diagnostic outcomes. DCA further demonstrated significant clinical net benefit. In film-reading comparison experiments, the independent diagnosis group of physicians achieved an accuracy of 0.9085 and recall rate of 0.8929, which were both lower than those of the model-only diagnosis group (accuracy 0.9306, recall 0.9071). The AI-assisted physician group achieved the highest diagnostic performance with an accuracy of 0.9590 and recall rate of 0.9429. Conclusion AI radiomics model can significantly enhance the accuracy and efficiency of X-ray diagnosis for tibial plateau fractures, effectively reducing the risks of missed and misdiagnoses. This provides reliable technical support for early precise diagnosis of fractures. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
