基于CT影像组学诺模图预测微小甲状腺结节良恶性

A radiomics nomogram based on computed tomography for predicting benign and malignant thyroid nodules

目的探讨基于CT的影像组学诺模图对≤1 cm甲状腺结节良恶性的预测价值。方法回顾性收集2019年1月至8月烟台毓璜顶医院收治的160例符合条件的甲状腺结节患者(良性56例、恶性104例)的影像学及临床资料,随机将其分为训练集(n=127)和验证集(n=33)。从患者的平扫和增强CT动脉期图像中分别提取影像组学特征。在训练集中,采用单因素方差分析(ANOVA)和最小绝对收缩和选择算子(LASSO)筛选与良恶性结节相关影像组学特征并构建影像组学标签,结合所选特征与其加权系数乘积的线性组合生成影像组学标签得分(影像组学评分)。采用ANOVA筛选独立临床危险因素,并采用多因素Logistic回归结合影像组学评分筛选最终预测因素构建影像组学诺模图。使用受试者工作特性(ROC)曲线评价模型预测效能。结果19个与甲状腺结节状态相关的特征组成的影像组学标签取得了良好的预测效果,并计算影像组学评分。多因素Logistic回归结果显示,甲状腺影像报告和数据系统(TI-RADS分级)、影像组学评分为甲状腺结节良恶性预测相关的独立危险因素。纳入这2种因素的影像组学诺模图在训练集(AUC:0.835;95%置信区间[CI]:0.776,0.884)和验证集中(AUC:0.793;95%CI:0.642,0.901)均显示出较好的鉴别能力。结论所提出的影像组学诺模图是一种结合影像组学特征和临床危险因素的无创预测工具,对≤1 cm甲状腺结节良恶性预测具有较高效能,显著优于常规影像学方法。

Objective To investigate the value of computed tomography(CT)radiomics nomogram for predicting benign and malignant thyroid nodules of sub-1 cm.Methods The imaging and clinical data of 160 eligible patients with thyroid nodules(56 benign and 104 malignant)at Yantai Yuhuangding Hospital between January and August 2019 were retrospectively collected and randomly divided into a training set(n=127)and a validation set(n=33).The radiomics features were extracted from the unenhanced and arterial contrast-enhanced CT images.In the training set,one-way analysis of variance(ANOVA)and the least absolute shrinkage and selection operator(LASSO)regression were used to select the features related to benign and malignant nodules,and radiomics signature score(Rad-score)was generated using a linear combination of the selected features weighted by the LASSO algorithm.ANOVA was used to select independent clinical risk factors and multivariate logistic regression combined with Rad-score to select the final predictors and construct a radiomics nomogram.The predictive ability of the radiomics nomogram was evaluated using the area under the curve(AUC)of receiver operating characteristic(ROC)analysis.Results The radiomics signature,consisting of 19 thyroid nodules-status-related features,achieved favorable prediction efficacy and calculated Rad-score.Multivariate logistic regression results showed that Thyroid Imaging Reporting and Data System(TI-RADS)and Rad-score were independent risk factors related to the prediction of benign and malignant thyroid nodules.The radiomics nomogram including the two factors showed high discrimination ability in the training set(AUC:0.835;95%confidence interval[CI]:0.776,0.884)and the validation set(AUC:0.793;95%CI:0.642,0.901).Conclusion The radiomics nomogram proposed in this study is a non-invasive prediction tool combining radiomics features and clinical risk factors,which has high efficiency and significantly superior to conventional imaging examination for predicting benign and malignant thyroid nodules of sub-1 cm.