摘要
目的 探讨基于增强CT影像组学联合临床的列线图在术前预测肾透明细胞癌(ccRCC)WHO/ISUP分级中的价值。方法 回顾性纳入214例经病理证实且术前行增强CT扫描的ccRCC患者。根据WHO/ISUP分级将其划分为低级别(Ⅰ~Ⅱ级)和高级别(Ⅲ~Ⅳ级),按照4∶1随机分为训练集和测试集。从平扫及三期增强图像中分割感兴趣区(ROI),提取影像组学特征,Spearman等级相关系数和LASSO回归进行特征筛选与降维,KNN算法构建影像组学模型;单因素及多因素分析筛选临床及影像语义学特征,KNN算法构建临床模型;结合临床及影像组学标签构建联合模型,并绘制列线图。ROC曲线和Delong检验评估模型的诊断性能,校准曲线和决策分析曲线评估模型的准确性和临床实用价值。结果 最终筛选出8个临床特征和11个影像组学特征。联合模型在训练集和测试集中的AUC值分别为0.887和0.800,优于临床模型和影像组学模型。校准曲线结果表明联合模型与真实结果之间具有良好的一致性,决策曲线分析结果表明列线图可以获得良好的净收益。结论 联合影像组学和临床标签构建的列线图可以为术前预测ccRCC分级提供证据,从而指导临床决策。
Objective To explore the utility of a nomogram integrating contrast-enhanced CT radiomics with clinical features in the preoperative prediction of WHO/ISUP grade for clear cell renal cell carcinoma(ccRCC).Methods A total of 214 patients with pathologically proven ccRCC who underwent enhanced CT scan before surgery were retrospectively included.According to the WHO/ISUP grade system,the cases were classified into low-grade(gradesⅠ-Ⅱ)and high-grade(gradesⅢ-Ⅳ),and then randomly divided into training and test set with a ratio of 4∶1.Regions of interest were segmented from both unenhanced and three-phase enhanced images,and radiomic features were extracted.Feature selection and dimensionality reduction were performed using Spearman rank correlation coefficients and LASSO regression,followed by the construction of the radiomic model with the KNN algorithm.Clinical and semantic imaging features were selected through univariate and multivariate analyses,and a clinical model was developed using the KNN algorithm.The clinical and radiomics signatures were used to construct a combined model and a nomogram was developed.The ROC curve and delong test were used to evaluate the diagnostic performance of the model,while calibration and decision curve analyses assessed its accuracy and clinical applicability.Results 8 clinical features and 11 radiomic features were selected.The combined model,integrating these clinical and radiomics signatures,exhibited robust predictive performance with AUC values of 0.887 in the training set and 0.800 in the test set.The calibration curve demonstrated good consistency between the nomogram model and actual outcomes,while decision curve analysis indicated a favorable net benefit for the nomogram.Conclusion The nomogram constructed by combining radiomics and clinical signatures can provide evidence for preoperative prediction of ccRCC grade and guide clinical decision-making.
作者
朱宏庆
张涛
顾康琛
王弦
管松
严彦
姚文君
Zhu Hongqing;Zhang Tao;Gu Kangchen;Wang Xian;Guan Song;Yan Yan;Yao Wenjun(Dept of Radiology,The Second Affiliated Hospital of Anhui Medical University,Hefei 230601;Dept of Urology,The Second Affiliated Hospital of Anhui Medical University,Hefei 230601;Dept of Pathology,The Second Affiliated Hospital of Anhui Medical University,Hefei 230601)
出处
《安徽医科大学学报》
北大核心
2025年第6期1127-1133,共7页
Acta Universitatis Medicinalis Anhui
基金
安徽省自然科学基金(编号:2008085MH290)
安徽省高校科研项目(编号:2024AH050796)
安徽省转化医学研究院科研基金(编号:2021zhyx-C45)
安徽医科大学第二附属医院临床研究培育计划(编号:2020LCYB05)。
作者简介
朱宏庆,男,硕士研究生;通信作者:姚文君,副教授,副主任医师,硕士生导师,E-mail:979839187@qq.com。
