基于CT的影像组学与深度学习用于术前甲状腺结节分类:系统综述、Meta分析与放射科医师比较
该系统综述与Meta分析检索PubMed、Embase、Scopus和Web of Science至2025年6月2日,纳入12项影像组学研究与5项深度学习研究,评估CT影像组学与深度学习模型术前分类甲状腺结节的诊断效能;影像组学模型汇总敏感度0.85、特异度0.83、AUC 0.894,深度学习模型汇总敏感度0.87、特异度0.93、AUC 0.911,模型整体表现优于放射科医师,平扫CT模型表现优于动脉期或静脉期模型。
要点
- 深度学习模型汇总特异度0.93、AUC 0.911,影像组学模型汇总特异度0.83、AUC 0.894。
- 模型整体表现优于放射科医师,平扫CT模型表现优于动脉期或静脉期模型。
- 影像组学模型异质性低,深度学习模型异质性较大,可能源于验证、分割、METRICS质量及参考标准差异。
核心结论
- CT影像组学与深度学习模型在甲状腺结节分类中表现出有前景的性能,可能提高放射科医师对不确定病例的准确率,同时减少不必要的活检。
研究设计与样本
- 系统综述与Meta分析,检索PubMed、Embase、Scopus和Web of Science自建库至2025年6月2日,纳入12项影像组学研究和5项深度学习研究,使用QUADAS-2和METRICS评估研究质量,采用双变量Meta分析汇总诊断效能。
诊断标准与关键数值
- 影像组学模型(12项研究):汇总敏感度0.85,特异度0.83,阳性似然比4.60,阴性似然比0.19,诊断比值比30.29,AUC 0.894。
- 深度学习模型(5项研究):汇总敏感度0.87,特异度0.93,阳性似然比14.04,阴性似然比0.15,诊断比值比95.76,AUC 0.911。
- 模型整体表现优于放射科医师,平扫CT图像模型表现优于基于动脉期或静脉期的模型。
临床应用与阅片要点
- CT影像组学与深度学习模型可能提高放射科医师对不确定甲状腺结节的分类准确率,并可能减少不必要的活检。
- 平扫CT图像模型表现优于动脉期或静脉期模型,阅片时可优先关注平扫CT的定量分析价值。
- 深度学习模型异质性较大,可能源于验证、分割、METRICS质量及参考标准差异,解读其汇总性能时需结合这些因素。
局限性
- 深度学习模型异质性较大,可能源于验证、分割、METRICS质量及参考标准差异。
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
INTRODUCTION: Computed tomography (CT) can evaluate thyroid cancer invasion into adjacent structures and is useful in identifying incidental thyroid nodules. Computer-aided diagnostic approaches may provide valuable clinical advantages in this domain. Here, we aim to evaluate the diagnostic performance of radiomics and deep-learning methods using CT imaging for preoperative nodule classification.
METHODS: A comprehensive search of PubMed, Embase, Scopus, and Web of Science was conducted from inception to June 2, 2025. Study quality was assessed using QUADAS-2 and METRICS. A bivariate meta-analysis estimated pooled sensitivity, specificity, positive and negative likelihood ratios (PLR and NLR), diagnostic odds ratio (DOR), and area under the curve (AUC). Two supplementary analyses compared AI model performance with radiologists and assessed diagnostic utility across CT imaging phases (plain, venous, arterial). Subgroup and sensitivity analyses explored sources of heterogeneity. Publication bias was evaluated using Deek's funnel plot.
RESULTS: The meta-analysis included 12 radiomics studies (sensitivity: 0.85, specificity: 0.83, PLR: 4.60, NLR: 0.19, DOR: 30.29, AUC: 0.894) and five deep-learning studies (sensitivity: 0.87, specificity: 0.93, PLR: 14.04, NLR: 0.15, DOR: 95.76, AUC: 0.911). Radiomics models showed low heterogeneity, while deep-learning models showed substantial heterogeneity, potentially due to differences in validation, segmentation, METRICS quality, and reference standards. Overall, these models outperformed radiologists, and models using plain CT images outperformed those based on arterial or venous phases.
CONCLUSION: Radiomics and deep-learning models have demonstrated promising performance in classifying thyroid nodules and may improve radiologists' accuracy in indeterminate cases, while reducing unnecessary biopsies.
引用
- Broomand Lomer N, Ahmadzadeh AM, Ashoobi MA, et al. CT-Based Radiomics and Deep Learning for Preoperative Thyroid Nodule Classification: A Systematic Review, Meta-analysis, and Radiologist Comparison. Acad Radiol. 2025. PMID: 41107124. doi:10.1016/j.acra.2025.09.045
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。