论文部分内容阅读
目的:探讨CT图像特征联合双能CT定量参数对甲状腺乳头状癌（PTC）颈部淋巴结转移的诊断价值。方法:回顾性分析2017年6月至2019年6月于南京医科大学第一附属医院经手术病理证实的103例PTC患者的术前双能CT图像。参考2002年美国头颈外科协会的淋巴结分区标准，将颈部淋巴结分为7区，采用淋巴结影像病理亚区对照的方法，共纳入245枚颈部淋巴结，根据病理结果，107枚淋巴结归为转移淋巴结组，138枚归为非转移淋巴结组。纳入分析的CT图像特征包括淋巴结大小、形状、边缘、强化程度、强化方式、钙化、囊变和结外侵犯。测量动静脉期双能CT定量参数包括标准化碘浓度（NIC）、标准化有效原子序数（Zn eff-c）和能谱曲线斜率（λn HU）。采用χ2检验或Mann-Whitney n U检验比较转移和非转移淋巴结组间CT图像特征和定量参数差异。分别基于CT图像特征、双能CT定量参数及两者联合构建logistic回归模型，并采用ROC曲线评估其诊断效能。n 结果:转移和非转移淋巴结组间CT图像特征差异均有统计学意义（n P<0.05）。转移淋巴结的动静脉期NIC、Zn eff-c及λn HU均高于非转移淋巴结（n P<0.001）。联合CT图像特征和双能CT定量参数logistic模型诊断PTC颈部淋巴结转移的曲线下面积为0.922，灵敏度为86.0%，特异度为92.8%。双能CT定量参数模型曲线下面积为0.912，灵敏度为84.1%，特异度为93.5%。两者的诊断效能均优于CT图像特征模型，后者的曲线下面积为0.783，灵敏度为71.0%，特异度为73.9%（n Z=5.212、4.554，n P均<0.001）。n 结论:相较于CT图像特征，双能CT定量参数对PTC颈部淋巴结转移的诊断价值更高，两者联合能进一步提高诊断准确性。“,”Objective:To evaluate the diagnostic value of the combination of CT image features and quantitative dual-energy CT (DECT) parameters in diagnosing cervical lymph nodes (LNs) metastasis from papillary thyroid carcinoma (PTC).Methods:Preoperative DECT imaging data of 103 patients with pathologically proven PTC in the First Affiliated Hospital of Nanjing Medical University from June 2017 to June 2019 were retrospectively analyzed. Taking 2002 American Association of Head and Neck Surgery criteria for LNs division as a reference, cervical LNs were divided into 7 levels. A total of 245 LNs were enrolled using radiological-pathological subzone comparison method. According to pathological results, 107 LNs were classified as metastatic LNs group and 138 LNs were classified as non-metastatic LNs group. CT image features including size, shape, margin, degree of enhancement, pattern of enhancement, calcification, cystic change and extra nodal extension were assessed. Quantitative DECT parameters including standardized iodine concentration (NIC), standardized effective atomic number (Zn eff-c) and slope of energy spectrum curve (λn HU) were calculated. The χn 2 test or Mann-Whitney n U rank sum test were used to compare the difference of CT image features and quantitative parameters between the two groups. The multivariate logistic regression analysis was used to build models based on CT image features, quantitative DECT parameters and their combination. The ROC curve analyses were performed to evaluate the diagnostic efficiency.n Results:Significant differences were found in all CT image features between metastatic and non-metastatic LNs groups (n P<0.05). All the arterial and venous phase DECT parameters in metastatic LNs group were higher than those in non-metastatic LNs group (n P<0.001). The area under curve (AUC), sensitivity, specificity of combination of logistic model based on both CT image features and quantitative DECT parameters were 0.922, 86.0% and 92.8% for diagnosing cervical LNs metastasis from PTC. The AUC value, sensitivity, specificity of logistic model based on quantitative DECT parameters were 0.912, 84.1% and 93.5%. Both of them outperformed logistic model based on CT image features with an AUC of 0.783, a sensitivity of 71.0% and a specificity of 73.9% (n Z=5.212, 4.554, n P<0.001).n Conclusions:Compared with CT image features, quantitative DECT parameters has better diagnostic performance in differentiating metastatic from non-metastatic LNs in patients with PTC. Integrating CT image features with quantitative dual-energy CT parameters together can furthermore improve the differentiating performance.