改进支持向量机分类方法及其在原发性肝癌筛查中的应用

为了尽早检测出原发性肝癌患者,辅助医生进行医疗决策,提高患者临床疗效,提出一种基于临床常规检验指标的筛查方法。该方法使用支持向量机建模,采用差异进化算法进行参数优化,以接收者操作特征曲线下面积的值作为模型评价测度,将得到的最优模型用于检验数据识别,以判断该数据属于健康、良性病变还是原发性肝癌。此外,还根据临床需求绘制分类模型的性能指标曲线和阈值查找表,由用户选择阈值,使预测性能进一步提升。实验结果表明:与其他5种分类方法相比,该方法建立的模型具有更好的性能,其准确度可达0.94,Kappa系数可达0.90。研究结果可辅助医生进行原发性肝癌早期筛查,提高患者长期生存率。

Detection of PRL-2 gene expression in hepatocellular carcinoma by real-time fluorescence quantitative PCR

Objective:To quantitatively detect the expression level of PRL-2 in primary hepatocellular carcinoma using real-time fluorescence quantitative PCR.Methods:Total RNA isolated from human HCC and liver tissue adjacent to the tumor was reversely transcribed into cDNA.Real-time fluorescence quantitative PCR(Q-PCR) method was used to analyze the expres-sion level of PRL-2 gene.Results:The Q-PCR method was performed successfully to precisely detect RNA level.PRL-2 was expressed in all portal vein tumor thrombosis(PVTT) and HCC,but only in some paratumor tissue.The highest expression level of PRL-2 gene was recorded in PVTT;meanwhile expression level of PRL-2 was higher than that in paratumor liver tis-sues and in HCC(P < 0.01),and it was higher in HCC than that in paratumor liver tissues.Conclusion:The Q-PCR may be the most precise method to quantitatively detect RNA level and can be used in gene expression changes.The PRL-2 gene has higher expression in PVTT than that in HCC and in paratumor liver tissue cells,indicating that it plays an important role in the development and metastasis of the HCC.