Abstract:

Objective　To explore the value of spectral CT combined with ultrasound C-TIRADS grading in the differentiation of benign and malignant thyroid nodules. Methods　Seventy patients with thyroid nodules treated at Binzhou Medical University Hospital from May 2023 to May 2024 were selected for the retrospective study, including 26 cases of benign nodules and 44 cases of malignant nodules. All the patients took ultrasound examination and spectral CT scanning. The ages, genders, maximum nodule diameters, and spectral CT parameters were compared between the two groups. The independent predictive factors of spectral CT were selected   through univariate and multivariate analyses, and ultrasound C-TIRADS grading was introduced to construct the nomogram model. The model's stability was internally validated by the Bootstrap method with 1 000 iterations. The independent-sample t test, Mann-Whitney U test, and χ² test were used for the statistical analysis. Results　There were statistical differences in the CT spectral parameters, including iodine concentration (IC) in the arterial and venous phases, normal iodine concentration (NIC), and slope of the energy spectrum curve (λHU), as well as the maximum nodule diameter between the benign group and the malignant group (all P<0.05). The multivariate analysis revealed that arterial phase IC and venous phase NIC were independent predictive factors for determining the benign or malignant thyroid nodules (both P<0.05). A predictive model was constructed based on these variables, and its area under the curve (AUC) was 0.940. Utilizing ultrasound C-TIRADS grading for diagnosing benign or malignant thyroid nodules resulted in an AUC of 0.823. When ultrasound C-TIRADS grading was combined with spectral CT parameters to create a nomogram, its AUC was 0.982. The calibration curve demonstrated excellent calibration performance for the nomogram, with a Brier score of 0.051. The decision curve analysis indicated that the nomograms consistently provided good clinical net benefit across a wide range of threshold probabilities. The internal validation of the nomogram model using the Bootstrap method with 1 000 iterations further confirmed its robustness, yielding an average AUC of 0.961. Conclusions　The AUC of the spectral CT predictive model is higher than that of the ultrasound C-TIRADS grading. The combined model improves the performance of both spectral CT and ultrasound C-TIRADS grading in distinguishing benign and malignant thyroid nodules.

Key words:

Thyroid nodules, Spectral CT, Ultrasound C-TIRADS grading, Diagnosis

摘要：

目的　探讨能谱CT联合超声C-TIRADS分级鉴别甲状腺结节良恶性的价值。方法　选取2023年5月至2024年5月滨州医学院附属医院收治的70例甲状腺结节患者进行回顾性分析，其中良性结节26例，恶性结节44例。术前均行超声检查及能谱CT增强扫描。比较两组年龄、性别、结节长径、能谱CT参数等资料。通过单因素及多因素分析筛选出能谱 CT 的独立预测因素， 引入 超声 C-TIRADS 分级构建列线图模型。采用Bootstrap法迭代1 000次，内部验证模型的稳定性。采用独立样本t检验、Mann-Whitney U检验、χ²检验进行统计分析。结果　良性组和恶性组能谱参数［包括动脉期及静脉期碘浓度（iodine concentration，IC）、标准化碘浓度（normal iodine concentration，NIC）、能谱曲线斜率（slope of the energy spectrum curve，λHU）］以及结节长径比较，差异均有统计学意义（均P<0.05）。多因素分析表明，动脉期IC及静脉期NIC是鉴别甲状腺结节良恶性的独立预测因素（均P<0.05）。基于上述变量构建预测模型，该模型曲线下面积（AUC）为0.940。利用超声C-TIRADS分级诊断甲状腺结节良恶性，其AUC为0.823。超声C-TIRADS分级联合能谱CT参数构建列线图，其AUC为0.982。校准曲线显示，列线图校准度表现优秀，Brier评分为0.051。决定曲线分析显示，在广泛阈值概率范围内，列线图均表现出较好的临床净收益。应用Bootstrap法进行1 000次迭代，计算平均AUC来对列线图模型进行内部验证，平均AUC为0.961。结论　能谱CT预测模型AUC高于超声C-TIRADS分级。联合模型可以提高能谱CT及超声C-TIRADS分级鉴别甲状腺结节良恶性的效能。

关键词:

甲状腺结节, 能谱CT, 超声C-TIRADS分级, 诊断