期刊文献+

影像组学在局部晚期鼻咽癌化疗敏感性及预后中的应用 被引量:3

Application of Imageomics in Chemosensitivity and Prognosis of Locally Advanced Nasopharyngeal Carcinoma
原文传递
导出
摘要 目的 探索治疗前MRI影像组学特征在预测局部晚期鼻咽癌化疗后肿块残留及临床预后的应用价值。方法 回顾性分析167例病理Ⅱ型、临床分期为Ⅲ~Ⅳb期鼻咽癌初治患者,接受2~3周期诱导化疗及放疗,同时具有治疗前2周内、化疗后一周内T2WI抑脂(FS)和FS T1WI增强(CE)两个序列图像。根据化疗后复查MR图像肿瘤是否有残留分为肿块残留组(95例)和完全回缩组(72例)。分别勾画治疗前两个序列图像中鼻咽癌肿块作为感兴趣区(ROI)进行特征提取,首先采用最大相关最小冗余(mRMR)以消除冗余和无关特征,然后,使用LASSO选择优化的特征子集来构建模型,使用受试者工作特征(ROC)曲线评估模型的预测效能。最后运用Kaplan-Meier法以无进展生存率(PFS)绘制出生存曲线。结果 经过特征计算和模型的训练验证,在训练队列及验证队列中的曲线下面积(AUC)、准确率、灵敏度、特异度、阴性预测值和阳性预测值分别为0.82和0.77、74.58%和71.43%、71.64%和78.57%、78.43%和61.90%、67.80%和68.42%、81.36%和73.33%;生存分析显示肿块完全回缩组和肿块残留组5年PFS为76.83%vs.43.58%(P<0.05)。结论 治疗前MRI影像组学特征在预测局部晚期鼻咽癌化疗后肿块是否残留及治疗后临床有无进展有一定预测价值,可指导鼻咽癌的个体化治疗。 Objective To explore the value of pretreatment MRI radiomics in predicting residual tumor and clinical prognosis after chemotherapy for locally advanced nasopharyngeal carcinoma.Methods A total of 167 treatment-naive patients with pathological typeⅡand clinical stages Ⅲ-Ⅳb nasopharyngeal carcinoma who received 2-3 cycles of induction chemotherapy and radiotherapy were retrospectively analyzed,there were also two sequence images,T_(2)WI FS and FS T_(1)WI CE,which within 2 weeks before treatment and within one week after the end of chemotherapy.They were divided into residual tumor group(95 cases)and complete retraction group(72 cases)according to whether there was residual tumor on reexamination of magnetic resonance images after chemotherapy.The nasopharyngeal carcinoma mass in the two sequence images was delineated separately as a region of interest for feature extraction.Firstly,used maximum correlated minimum redundancy(mRMR)to eliminate redundant and irrelevant features,then the optimized feature subset was selected using LASSO to build the model,the receiver operating characteristic(ROC)was used to assess the predictive performance of the model.Finally,survival curves for progression-free survival were described using the Kaplan-Meier method.Results After feature calculation and model training validation,the AUC、Accuracy、Sensitivity、Specificity、Positive predictive value and Negative predictive value in the training cohort and validation cohort were 0.82 and 0.77,74.58%and 71.43%,71.64%and 78.57%,78.43% and 61.90%,81.36% and 73.33%,67.80% and 68.42%respectively;survival analysis showed that the 5-year PFS was 76.83%vs.43.58%in the complete shrinkage group and the residual mass group,P<0.05.Conclusion Pretreatment MRI radiomics features have some predictive value in predicting whether the mass remains after chemotherapy for locally advanced nasopharyngeal carcinoma and whether there is clinical progression after treatment,which can guide the individualized treatment of nasopharyngeal carcinoma.
作者 纪海明 郗玉珍 张濬涛 丁忠祥 JI Haiming;XI Yuzhen;ZHANG Juntao(Department of Radiology,Hangzhou First People's Hospital,Zhejiang University School of Medicine,Hangzhou,Zhejiang Province 310006,P.R.China)
出处 《临床放射学杂志》 北大核心 2022年第10期1840-1845,共6页 Journal of Clinical Radiology
基金 国家自然科学基金项目资助(编号:81871337) 杭州市医药卫生科技项目(编号:B20220177)。
关键词 鼻咽肿瘤 影像组学 化疗 敏感性 生存分析选择 Nasopharyngeal neoplasms Radiomics Chemotherapy Sensitivity Survival analysis selection
  • 相关文献

参考文献2

二级参考文献8

共引文献23

同被引文献38

引证文献3

二级引证文献2

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部