目的：分析8种血清及血常规指标在原发性肝癌和肝炎患者鉴别诊断中的特异性，建立BP诊断模型。方法：采用回顾性研究的方法，收集经病理确诊的96例原发性肝癌患者为肝癌组，对应收集同期109例肝炎患者作为肝炎组。分别收集两组患者的高尔基体蛋白73（GP73）、甲胎蛋白（AFP）、α-L-岩藻糖苷酶（AFU）、谷草转氨酶（AST）、谷丙转氨酶（ ALT）、白细胞计数（ WBC）、红细胞计数（ RBC）和血小板计数（ PLT）等8种血清及血常规指标数据。利用ROC曲线分析各指标的敏感度，根据尤登指数计算两组界限值，然后建立纳入不同指标的BP神经网路模型，并计算模型的准确率。结果：GP73、AFP、AFU、AST、ALT、WBC、RBC、PLT 等8个指标的ROC曲线下面积分别为0．996、1．000、0．990、0．806、0．680、0．800、0．419和0．460。分别纳入8种指标和去掉RBC、PLT后的6种指标成功建立BP神经网络诊断模型，其准确率分别为86．0％（39／45）和95．5％（43／45）。结论：GP73、AFP、AFU、AST、ALT、WBC指标在肝癌和肝炎患者的鉴别诊断中具有重要意义，对于肝炎患者的病情进展具有一定的提示意义。%Aim:To analyze the specificity of eight serum and blood indexes in differential diagnosis of liver cancer and hepatitis,and establish a BP diagnostic model .Methods:A total of 96 primary liver cancer patients confirmed by pathology were enrolled as case group, 109 hepatitis patients were treated as control group .The data of Golgi protein 73(GP73),alpha fetal protein(AFP), alpha-L-fucosidase(AFU), aspartate transaminase(AST), glutamic pyruvic transaminase(ALT), white blood cell count ( WBC) , red blood cell count ( RBC) and blood platelet count ( PLT) in two group were collected respective-ly,and their specificity and sensibility according to the ROC curves were obtained .The BP network models were established with different indexes and the efficiency was evaluated .Results: The AUC of GP73,AFP,AFU,AST,ALT,WBC,RBC,and PLT were 0.996,1.000,0.990,0.806,0.680,0.800,0.419 and 0.460, respectively, and RBC and PLT were filtrated.The accuracy of the BP network models with all the indexes and the 6 indexes were 86.0%(39/45) and 95.5%(43/45).Con-clusion:The indexes GP73,AFP,AFU,AST,ALT and WBC have a relatively higher specificity in differential diagnosis of liv-er cancer and hepatitis , which may have certain implications in the progression of hepatitis patients .
展开▼
