蛋白质指纹图谱模型在肝癌诊断中的应用

目的:建立基于人工神经网络的血清蛋白质指纹图谱模型,并探讨其在肝癌诊断中的应用.方法:应用表面增强激光解析-电离-飞行时间质谱仪(SELDI-TOF-MS),测定了52例肝癌和32例健康人血清标本的蛋白质指纹图谱,并结合人工神经网络方法进行数据分析.将84例标本随机分成训练组56例(肝癌35例,健康人21例)和盲法测试组28例(肝癌17例,健康人11例).利用从训练组得出的基于人工神经网络的血清蛋白质指纹图谱模型,对28例未知血清进行检测,并与甲胎蛋白(AFP)判断结果进行比较.结果:应用该方法对肝癌进行诊断的准确率、敏感性和特异性均为100%(17/17,11/11).AFP检测结果准确率为77.4%(65/84),灵敏度和特异性分别为69.2%(36/52)和90.6%(29/32).结论:该方法在肝癌的诊断中较AFP具有更高的敏感性和特异性.

Application of protein fingerprint pattern models in liver cancer diagnosis

Aim: To evaluate the application of serum protein fingerprint pattern based on artificial neural network in diagnosis of liver cancer.Methods: A total of 52 cases of the patients with primary liver cancer and 32 cases of the healthy people were tested by WCX2 chip and proteinchip reader. A total of 84 samples were allocated into training group (n=56, 35 cases of liver cancer and 21 cases of healthy people) and blind test group (n=28, 17 cases of liver cancer and 11 cases of healthy people). The serum samples of the blind test group was detected using the protein fingerprint pattern model that obtained from the training group. The results were compared with those using AFP.Results: The accuracy, sensitivity, and specificity of using the potein fingerprint pattern were all 100%, and those of using AFP were 77.4%,69.2%, and 90.6%,respectively.Conclusion: Liver cancer can be quickly and exactly diagnosed by this method with higher sensitivity and specificity.