CT影像组学在预测鼻息肉术后疗效中的价值

The value of CT radiomics in predicting treatment outcomes in patients with nasal polyps

目的:探讨鼻窦CT影像组学在预测鼻息肉术后疗效中的价值。方法:采用回顾性队列研究,分析2015年1月至2019年12月于中山大学附属第一医院接受内镜鼻窦手术的194例鼻息肉患者(男132例,女62例,年龄16~75岁),随机法以约7∶3的比例将病例分为训练集135例和内部验证集59例;同时,纳入2020年1月至2021年12月中山大学附第七医院的34例鼻息肉患者(男22例,女12例,年龄16~59岁)作为外部验证集。按照2020年欧洲鼻窦炎和鼻息肉意见书(EPOS 2020)对病例进行术后疾病控制状态评估,并将病情未控制的病例定义为难治性鼻息肉。提取影像组学特征后,利用套索算法(least absolute shrinkage and selection operator,LASSO)筛选特征,根据加权系数建立影像组学模型;并构建临床特征预测模型和影像组学-临床特征联合模型;评估3类模型在预测难治性鼻息肉中的诊断性能。采用R软件(版本4.3.1)进行统计学分析。结果:在训练集,基于4个特征的影像组学和联合模型预测难治性鼻息肉的曲线下面积(AUC)分别为0.901和0.915,均高于传统临床特征模型(AUC=0.874);在内部验证集,3种模型的AUC值分别为0.839、0.832和0.716;外部验证集的3种模型预测AUC值有所下降,分别为0.748、0.764和0.620。所有队列中,影像组模型和联合模型预测难治性鼻息肉在决策曲线分析(decision curve analysis,DCA)中均显示了临床效益。结论:基于鼻窦CT的影像组学模型对难治性鼻息肉具有良好的预测效能,展现了潜在的临床应用价值。

Objective To evaluate the predictive efficacy of sinus CT radiomics for treatment outcomes in nasal polyp patients undergoing endoscopic sinus surgery.Methods A retrospective cohort study was conducted at the First Affiliated Hospital of Sun Yat-sen University,including 194 patients with nasal polyps treated between January 2015 and December 2019.The cohort comprised 132 males and 62 females,aged 16 to 75 years.Patients were divided into a training set(n=135)and an internal validation set(n=59).An external validation set(n=34),consisting of 22 males and 12 females aged 16 to 59 years,was included from January 2020 to December 2021.Disease control was evaluated using the criteria from the European Position Paper on Rhinosinusitis and Nasal Polyps 2020(EPOS 2020).Radiomic features were extracted from sinus CT images and analyzed using the least absolute shrinkage and selection operator(LASSO)regression.Models combining radiomic and clinical features were developed to predict treatment efficacy.Results The radiomics and combined models,based on four selected features,outperformed the clinical feature model in the training set,with AUC values of 0.901 and 0.915,versus 0.874,respectively.In the internal validation set,AUCs were 0.839,0.832,and 0.716.Despite reduced AUCs in the external set,the radiomics model maintained good generalizability(0.748,0.764,0.620).Decision curve analysis showed significant clinical benefits in both radiomics and combined models.Conclusion The CT-based radiomics model demonstrates significant predictive power in identifying refractory nasal polyps,suggesting its potential for clinical application in treatment outcome prediction.