CT纹理分析在非小细胞肺癌Ⅲ期CT引导下^(125)I粒子植入治疗中的疗效预测研究

Application of CT texture analysis in predicting the efficacy of ^(125)I particle implantation in stage Ⅲ non-small cell lung cancer

目的探讨CT纹理分析在非小细胞肺癌(NSCLS)Ⅲ期CT引导下经皮植入^(125)I粒子治疗中预测疗效的诊断价值与指导治疗方案临床应用价值研究。方法选取28例经穿刺活检病理证实的NSCLSⅢ期患者植入^(125)I粒子(联合化疗或在单独植入粒子治疗前3个月内接受过至少1种及以上药物治疗方案),在经临床评估为诊疗中前期(6个月内)非进展后,获取术前、术后3、6个月增强CT图像,手动勾画感兴趣区后提取术前与术后3个月、6个月随访CT图像的纹理特征。根据临床实际疗效对患者进行分组,比较各组间纹理参数。利用受试者工作(ROC)曲线下面积(AUC)、准确度、灵敏度、特异度评估模型,评价^(125)I粒子在NSCLSⅢ期中的^(125)I粒子治疗中疗效预测的诊断价值和指导治疗方案的临床的应用价值。结果在运用了纹理分析方法后对3个月期疗效获得2个纹理参数,对6个月期疗效获得6个纹理参数,通过纹理参数构建模型获取ROC曲线与AUC值,评估分组方法对疗效预测的诊断价值是积极的。结论结合CT纹理分析方法,在NSCLSⅢ期患者植入^(125)I粒子治疗时根据疗效的积极与否对患者群体进行分类有益于提供临床诊断价值和决策价值。

Objective To investigate the diagnostic value of CT texture analysis in predicting curative effect and guiding clinical application of treatment plan in stage Ⅲ non-small cell lung cancer(NSCLS) treated by CTguided percutaneous implantation of^(125)I seeds. Methods A total of 28 patients with stage Ⅲ NSCLS confirmed by biopsy and pathology were selected for implantation of^(125)I seeds(combined with chemotherapy or receiving at least one or more drug regiments within 3 months prior to implantation alone).After the patiens were clinically assessed as non-progressive in the middle and early stages of diagnosis and treatment(within 6 months), the enhanced CT images were obtained 3 and 6 months before and after the operation. After manually delineating the areas of interest, the texture features of the CT images were extracted 3 and 6 months before and after the operation. Patients were grouped according to clinical practice, texture parameters among groups were analyzed by multi-factor modeling. Using receiver operating characteristic(ROC) area under curve(AUC), accuracy, sensitivity and specificity evaluation model, the diagnostic value of^(125)I seeds in efficacy prediction and clinical application value of guiding treatment plan in stage Ⅲ NSCLS was evaluated. Results After applying the texture analysis method, two texture parameters were obtained for the 3-month efficacy and six texture parameters were obtained for the 6-month efficacy. ROC curve and AUC values were obtained by constructing the model based on texture parameters. It was positive to evaluate the diagnostic value of grouping method for efficacy prediction. Conclusion Combined with CT texture analysis method, classification of patients with stage Ⅲ NSCLS treated with^(125)Iseeds according to whether the efficacy is positive or not is beneficial to provide certain clinical diagnostic value and clinical treatment plan decision-making value.