光谱CT碘密度图联合影像组学预测食管癌新辅助化疗联合免疫治疗病理完全缓解的价值OA
Radiomics analysis of DLSCT-derived iodine density maps for predicting the pathological complete response of esophageal cancer after nCIT
目的:探讨基于光谱 CT 碘密度图的影像组学模型对食管癌新辅助化疗联合免疫治疗(nCIT)后病理完全缓解(pCR)的预测价值.方法:回顾性纳入 2022 年 10 月—2025 年 1 月在郑州大学第一附属医院接受 nCIT 并手术的 129 例食管鳞癌患者的临床病理及影像资料,根据肿瘤退缩分级(TRG)分为 pCR 组和非 pCR 组.按 7∶3 比例将患者随机划分为训练集(n=90)与验证集(n=39).所有患者于 nCIT 前接受光谱 CT 增强扫描,测量并计算碘密度、标准化碘密度、有效原子序数及光谱曲线斜率等光谱参数.基于静脉期碘密度图像勾画感兴趣区并提取影像组学特征,使用特征相关性分析结合 F 检验进行特征筛选,获取最优特征子集.应用单因素及多因素 Logistic 回归确定 pCR 的独立预测因子,并基于线性判别分析构建临床模型、光谱模型、组学模型和联合模型.采用受试者工作特征(ROC)曲线及分类准确度评估各模型预测效能,使用五折交叉验证对训练集各模型进行测试,并在验证集中进行评估.结果:多因素 Logistic 回归分析显示分化程度、临床 N 分期及动、静脉期的标准化碘密度(NIDA、NIDV)为 pCR 的独立预测因子.在训练集中临床模型、光谱模型、组学模型及联合模型的平均 AUC值分别为 0.66±0.10、0.66±0.07、0.67±0.04 和 0.85±0.09,平均分类准确度分别为 0.69±0.03、0.72±0.08、0.70±0.06 和 0.88±0.07,联合模型具有最优的诊断效能.在验证集中联合模型的 AUC 值为 0.88,分类准确度为 0.79.结论:基于光谱 CT 碘密度图影像组学特征联合和临床病理特征及光谱参数构建的联合模型,在预测食管癌 nCIT 后 pCR 方面表现出良好的性能,有望为个体化治疗决策提供影像学依据.
Objective:To investigate the value of radiomics analysis of dual-layer spectral detector CT(DLSCT)-derived iodine maps for evaluating the efficacy of neoadjuvant chemo-immunotherapy(nCIT)for esophageal cancer.Methods:This ret-rospective study included 129 patients with pathologically confirmed esophageal squamous cell carcinoma(ESCC)who under-went nCIT followed by surgery at the First Affiliated Hospital of Zhengzhou University between October 2022 and January 2025.Patients were categorized into pCR and non-pCR groups based on tumor regression grade(TRG).They were randomly divided into a training set(n=90)and a validation set(n=39)in a 7∶3 ratio.All patients underwent enhanced spectral CT scan before nCIT,and spectral parameters including iodine density,normalized iodine density,effective atomic number,and spectral curve slope were measured and calculated.Regions of interest(ROI)were delineated on venous-phase iodine density maps for radiomic feature extraction.The feature correlation analysis and F-test were used to select the optimal features.Univariate and multivariate Logistic regression analyses were used to identify independent predictors of pCR.Clinical,spec-tral,radiomics,and combined models were constructed based on linear discriminant analysis.Model performance was evalu-ated using receiver operating characteristic(ROC)curves and classification accuracy.Five-fold cross-validation was applied to the training set,and the models were further assessed in the validation set.Results:Multivariate Logistic regression analy-sis showed that the degree of differentiation,clinical N stage,and normalized iodine density in the arterial and venous phases(NIDA,NIDV)as independent predictors of pCR.In the training cohort,the mean AUC values for the clinical,spec-tral,radiomics,and combined models were 0.66±0.10,0.66±0.07,0.67±0.04,and 0.85±0.09,respectively,with mean classifi-cation accuracies of 0.69±0.03,0.72±0.08,0.70±0.06,and 0.88±0.07,respectively.The combined model demonstrated the best diagnostic performance.In the validation set,the combined model achieved an AUC of 0.88 and a classification accu-racy of 0.79.Conclusions:The combined model,integrating radiomics features from DLSCT iodine density maps with clini-copathological characteristics and spectral parameters,shows favorable performance in predicting pCR after nCIT in ESCC,and has the potential to serve as a tool for individualized treatment decision-making.
慕宛伶;罗成龙;孟新华;李峰;齐宇;党晋晋;高剑波;周悦
郑州大学第一附属医院,河南 郑州 450052郑州大学第一附属医院,河南 郑州 450052郑州大学第一附属医院,河南 郑州 450052郑州大学第一附属医院,河南 郑州 450052郑州大学第一附属医院,河南 郑州 450052飞利浦临床科学部,北京 100176郑州大学第一附属医院,河南 郑州 450052郑州大学第一附属医院,河南 郑州 450052
医药卫生
食管肿瘤放化疗,辅助体层摄影术,螺旋计算机
Esophageal NeoplasmsChemoradiotherapy,AdjuvantTomography,Spiral Computed
《中国临床医学影像杂志》 2026 (8)
561-566,6
河南省科技攻关(252102310073)郑州市基础研究与应用基础研究项目(2024ZZJCYJ007).
评论