论文部分内容阅读
目的比较4种不同的示踪动力学模型使用动态对比增强CT数据分析来预测原发性结直肠癌病人5年总生存率。材料与方法本研究通过伦理审查委员会批准。将46例结直肠癌病人存档的动态增强CT资料作为此研究的一部分,使用分布参数、常规分区、绝热组织同质性和普遍的动力学模型进行回顾性分析。用Friedman检验对血流量、血容量、平均通过时间(MTT)、渗透性表面面积乘积、摄取分数、血管外细胞外容积(ve)和容积转换常数(Ktrans)进行比较,统计学显著性为5%。根据受试者操作特征分析,使用KaplanMeier分析和单变量Cox比例风险模型,并且附加交叉验证和置换检验比较不同动力学模型的参数和肿瘤分期来分析总生存率的差别。结果分布参数模型的血流量比常规分区模型和绝热组织同质性模型要低(P<0.0001),而且常规分区模型与绝热组织同质性模型的血流量值是相似的。相反,分布参数模型的MTT比常规分区模型和绝热组织同质性模型要长(P<0.0001)。血容量、渗透性表面面积乘积以及血管外细胞外容积,在常规分区模型中比在绝热组织同质性模型、分布参数模型或普遍的动力学模型中都要高(P<0.0001)。分布参数模型的摄取分数比绝热组织同质性模型要高。至于5年总生存率,在进行交叉验证和置换检验后,只有分布参数模型派生出的阈值为15.48mL/100mL的血管外细胞外容积可以预测5年总生存率。结论模型之间的参数值有显著差异。模型研究的结果表明分布参数模型对预测5年总生存率是最好的。
Objective To compare four different models of tracer kinetics using dynamic contrast-enhanced CT data analysis to predict the 5-year overall survival of patients with primary colorectal cancer. Materials and Methods This study was approved by the Ethics Review Board. Forty-six patients with colorectal cancer were enrolled in dynamic contrast-enhanced computed tomography (CT) data as part of this study. Retrospective analysis was performed using distribution parameters, routine zoning, adiabatic tissue homogeneity, and general kinetic models. Friedman’s test was used to compare blood flow volume, blood volume, mean transit time (MTT), product of osmotic surface area, uptake fraction, extravascular extracellular volume (ve) and volumetric transfer constant (Ktrans) with a statistical significance of 5 %. Kaplan Meier analysis and univariate Cox proportional hazards models were used and the differences in overall survival were analyzed using cross-validation and permutation tests comparing parameters of different kinetic models and tumor staging based on subject manipulation profiling. Results The distribution parameter model had lower blood flow than the normal and adiabatic tissue homogeneity models (P <0.0001), and the blood flow values for the conventional and adiabatic tissue homogeneity models were similar. In contrast, the MTT of the distribution parameter model was longer (P <0.0001) than the conventional partition model and adiabatic tissue homogeneity model. The product of blood volume, osmotic surface area, and extravascular extracellular volume was higher in the conventional zoned model than in the adiabatic tissue homogeneity model, the distribution parameter model or the general kinetic model (P <0.0001). The distribution parameter model has a higher intake fraction than the adiabatic tissue homogeneity model. As for the 5-year overall survival rate, only the extra-vascular extracellular volume with a threshold value of 15.48 mL / 100 mL derived from the distribution parameter model can predict 5-year overall survival after cross-validation and permutation tests. Conclusion The parameters of the model have significant differences. The results of the model studies show that the distributional parameter model is the best for predicting 5-year overall survival.