基于CT影像组学预测甲状腺乳头状癌甲状腺外侵犯的价值

The Value of CT-based Radiomics in Predicting Extrathyroidal Extension of Papillary Thyroid Carcinoma

目的:探究CT影像组学预测甲状腺乳头状癌(Papillary Thyroid Cancer,PTC)发生甲状腺外侵犯(Extrathyroidal Extension,ETE)的评估价值。方法:选取2021年10月至2022年10月于我院收治的PTC患者140例作为研究对象,采用随机数法按7∶3比例将患者分为训练集(n=98)和测试集(n=42),依据其是否发生ETE将训练集患者分为ETE组(n=55)和non-ETE组(n=43),基于CT影像组学评分及多元Logistic回归模型分析影响PTC患者发生ETE的关键因素,模型中纳入的变量基于单因素分析的结果,并采用逐步回归方法进行变量筛选。模型的拟合优度通过Hosmer-Lemeshow拟合优度检验进行评估。ROC曲线用于评估模型的区分能力,而校准曲线则用于评估模型的预测准确度。结果:两组患者在肿瘤边界、桥本甲状腺炎、中央区淋巴结转移、多灶性、肿瘤最大径、颈侧区淋巴结转移、年龄方面差异有统计学意义(均P<0.05);多元Logistic回归模型结果表明肿瘤最大径、多灶、颈部淋巴结转移、年龄、平扫Rad-score、增强扫描Rad-score是影响PTC患者发生ETE的独立危险因素;并基于CT影像组学评分和其独立危险因素绘制列线图,ROC曲线、校准曲线结果显示该预测模型区分度和准确度较好。训练集(AUC=0.962)和测试集(AUC=0.937)的AUC值均接近1,表明该预测模型具有良好的区分能力,能够准确地区分发生ETE和未发生ETE的PTC患者。这为临床医生在制定治疗方案和评估患者预后提供了重要的参考依据。结论:临床医师对于年龄≥55岁、肿瘤最大径>2.0 cm、伴有多灶、颈部淋巴结转移PTC患者,基于CT影像组学可较好地预测ETE发生风险,可为临床提供参考依据。

Objective:To evaluate the value of CT-based radiomics in predicting extrathyroidal extension(ETE)of papillary thyroid cancer(PTC).Methods:140 patients with PTC were selected as the study subjects who were admitted to our hospital from October 2021 to October 2022.The patients were randomly divided into training set(n=98)and test set(n=42)at a ratio of 7∶3.According to whether there was ETE,the patients were divided into ETE group(n=55)and non-ETE group(n=43).Based on CT imaging omics scores and multiple logistic regression models,the key factors affecting the occurrence of ETE in PTC patients were analyzed.The variables included in the model were based on the results of univariate analysis,and used stepwise regression method for variable screening.The goodness of fit of the model was evaluated through the Hosmer-Lemeshow goodness of fit test.The ROC curve was used to evaluate the model’s discriminative ability,while the calibration curve was used to evaluate the model’s predictive accuracy.Results:There were significant differences between the two groups in tumor border,age,maximum tumor diameter,multifocal,central lymph node metastasis,cervical lymph node metastasis,Hashimoto’s thyroiditis(P<0.05).Multiple Logistic regression model showed that maximum tumor diameter,age,multiple foci,cervical lymph node metastasis,plain scan Rad score and enhanced scan Rad score were independent risk factors for the occurrence of ETE in PTC patients.And based on the CT imaging score and clinical risk factors,a nomogram prediction model was constructed.The results of ROC curve and calibration curve showed that the prediction model had good discrimination and accuracy.The AUC values of the training set(AUC=0.962)and the test set(AUC=0.937)were both close to 1,indicating that the predictive model had good discriminative ability and could accurately distinguished PTC patients with and without ETE.This provided important reference for clinical doctors in formulating treatment plans and evaluaed patient prognosis.Conclusion:Based on CT radiomics,clinical physicians can better predict the risk of ETE in PTC patients aged≥55 years,with a maximum tumor diameter>2.0 cm,accompanied by multiple lesions and cervical lymph node metastasis,which can provide a reference basis for clinicians to choose preoperative surgical methods.