目的 采用基于增强CT(contrast-enhanced CT,CECT)的影像组学列线图鉴别术前胸腺瘤风险程度.方法 回顾性分析2018年1月-2023年2月在江苏省苏北人民医院行手术切除并经病理证实胸腺瘤患者的临床资料.从每例患者胸部CECT的动脉期中提取影像组学特征,采用Pearson相关系数、最小绝对收缩和选择算子(least absolute shrinkage and selection operator,LASSO)进行影像组学特性选择,极端随机树(extremely randomized trees,ExtraTrees)分类器构建影像组学特征模型和影像组学标签(rad signature).基于患者临床特征单因素、多因素logistic回归分析筛选出来的影像特征、临床特征构建临床特征模型.结合影像组学标签和临床特征构建影像组学列线图模型.计算曲线下面积(area under the curve,AUC)、敏感度、特异度、准确率、阴性预测值、阳性预测值指标评估模型性能,绘制校准曲线、决策曲线评估模型准确度和临床获益.结果 共纳入120例患者,其中女59例、男61例,平均年龄(56.30±12.10)岁.其中训练组84例,测试组36例;低风险胸腺瘤患者62例,高风险胸腺瘤患者58例.提取影像组学特征1 038个,从中筛选出6个影像组学特征用于建立影像组学标签.包含临床因素和影像组学标签的列线图模型在训练组中AUC为0.872,测试组中的AUC为0.833.决策曲线表明列线图模型在临床实用性方面优于影像组学标签和临床模型.结论 影像组学列线图对于胸腺瘤风险程度有较好的鉴别价值,为临床决策提供了一种无创、有效的方法.
Contrast-enhanced CT-based radiomics nomogram for differentiation of low-risk and high-risk thymomas
Objective To develop a radiomics nomogram based on contrast-enhanced CI(CECT)tor preoperative prediction of high-risk and low-risk thymomas.Methods Clinical data of patients with thymoma who underwent surgical resection and pathological confirmation at Northern Jiangsu People's Hospital from January 2018 to February 2023 were retrospectively analyzed.Feature selection was performed using the Pearson correlation coefficient and least absolute shrinkage and selection operator(LASSO)method.An ExtraTrees classifier was used to construct the radiomics signature model and the radiomics signature.Univariate and multivariable logistic regression was applied to analyze clinical-radiological characteristics and identify variables for developing a clinical model.The radiomics nomogram model was developed by combining the radiomics signature and clinical features.Model performance was evaluated using area under the curve(AUC),sensitivity,specificity,accuracy,negative predictive value,and positive predictive value.Calibration curves and decision curves were plotted to assess model accuracy and clinical values.Results A total of 120 patients including 59 females and 61 males with an average age of 56.30±12.10 years.There were 84 patients in the training group and 36 in the validation group,62 in the low-risk thymoma group and 58 in the high-risk thymoma group.Radiomics features(1 038 in total)were extracted from the arterial phase of CECT scans,among which 6 radiomics features were used to construct the radiomics signature.The radiomics nomogram model,combining clinical-radiological characteristics and the radiomics signature,achieved an AUC of 0.872 in the training group and 0.833 in the validation group.Decision curve analysis demonstrated better clinical efficacy of the radiomics nomogram than the radiomics signature and clinical model.Conclusion The radiomics nomogram based on CECT showed good diagnostic value in distinguishing high-risk and low-risk thymoma,which may provide a noninvasive and efficient method for clinical decision-making.
NETL
NSTL
万方数据
