文献速递头颈CT

基于CT深度学习诊断喉癌及下咽癌的甲状软骨侵犯

Diagnosis of thyroid cartilage invasion by laryngeal and hypopharyngeal cancers based on CT with deep learning.
PubMed 文献 · Eur J Radiol · 2025-05-13

回顾性分析91例经手术治疗的喉癌或下咽癌病例,按61例训练、30例测试划分数据集,以病理诊断为金标准,基于ResNet101迁移学习构建卷积神经网络模型,从治疗前轴位CT图像分类甲状软骨侵犯状态。测试集AUC为0.82,准确率90%、敏感度80%、特异度95%;模型AUC与低年资放射科医师有显著差异(p=0.035),与高年资医师无显著差异(p=0.61)。

要点

  • 测试集CNN模型诊断甲状软骨侵犯AUC 0.82,准确率90%、敏感度80%、特异度95%。
  • 模型AUC显著高于低年资医师(p=0.035),与高年资医师无显著差异(p=0.61)。
  • 作者认为该CNN模型可作为评估喉癌或下咽癌甲状软骨侵犯的辅助工具。

核心结论

  • 基于CNN的诊断模型可成为评估喉癌或下咽癌患者甲状软骨侵犯的有用辅助工具。

研究设计与样本

  • 回顾性分析2010年4月至2023年5月在本院经手术治疗的91例喉癌或下咽癌病例。
  • 数据集分为训练集61例和测试集30例。
  • 以全部病例的CT图像及病理诊断确定侵犯阳性或阴性状态作为金标准。
  • 基于ResNet101迁移学习训练新CNN模型,从治疗前轴位CT图像分类甲状软骨侵犯阳性或阴性状态。
  • 由两名放射科医师(一名头颈影像经验丰富的高年资医师、一名经验较少的低年资医师)审阅测试集CT图像判断有无甲状软骨侵犯。

诊断标准与关键数值

  • 测试集CNN模型AUC为0.82,准确率90%、敏感度80%、特异度95%。
  • CNN模型与低年资医师的AUC有显著差异(p=0.035)。
  • CNN模型与高年资医师的AUC无显著差异(p=0.61)。

临床应用与阅片要点

  • 作者认为该CNN模型可作为评估喉癌或下咽癌甲状软骨侵犯的辅助工具。

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

OBJECTIVES: To develop a convolutional neural network (CNN) model to diagnose thyroid cartilage invasion by laryngeal and hypopharyngeal cancers observed on computed tomography (CT) images and evaluate the model's diagnostic performance.

METHODS: We retrospectively analyzed 91 cases of laryngeal or hypopharyngeal cancer treated surgically at our hospital during the period April 2010 through May 2023, and we divided the cases into datasets for training (n = 61) and testing (n = 30). We reviewed the CT images and pathological diagnoses in all cases to determine the invasion positive- or negative-status as a ground truth. We trained the new CNN model to classify thyroid cartilage invasion-positive or -negative status from the pre-treatment axial CT images by transfer learning from Residual Network 101 (ResNet101), using the training dataset. We then used the test dataset to evaluate the model's performance. Two radiologists, one with extensive head and neck imaging experience (senior reader) and the other with less experience (junior reader) reviewed the CT images of the test dataset to determine whether thyroid cartilage invasion was present.

RESULTS: The following were obtained by the CNN model with the test dataset: area under the curve (AUC), 0.82; 90 % accuracy, 80 % sensitivity, and 95 % specificity. The CNN model showed a significant difference in AUCs compared to the junior reader (p = 0.035) but not the senior reader (p = 0.61).

CONCLUSIONS: The CNN-based diagnostic model can be a useful supportive tool for the assessment of thyroid cartilage invasion in patients with laryngeal or hypopharyngeal cancer.

来源:Eur J Radiol,查看原文。版权归原作者所有。

引用

  1. Takano Y, Fujima N, Nakagawa J, et al. Diagnosis of thyroid cartilage invasion by laryngeal and hypopharyngeal cancers based on CT with deep learning. Eur J Radiol. 2025. PMID: 40381388. doi:10.1016/j.ejrad.2025.112168

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。