血清蛋白质谱与人工神经网络模型诊断卵巢癌的应用性研究  

Classification and diagnostic prediction of ovarian cancers using protein profiling of serum and artificial neural networks

作　　者：余捷凯[1] 郑树[1] 唐勇[2] 李力[2] 胡跃[1] 刘建[1] 

机构地区：[1]浙江大学医学院附属第二医院肿瘤研究所,杭州310009 [2]广西医科大学附属肿瘤医院妇科

出　　处：《中华检验医学杂志》2005年第5期480-482,共3页Chinese Journal of Laboratory Medicine

基　　金：973 国家重点基础研究发展规划资助项目(G1998051200)

摘　　要：目的　建立筛选卵巢癌血清蛋白质谱与人工神经网络诊断模型的研究。方法　用H4(疏水表面)蛋白芯片结合表面增强激光解吸电离飞行时间质谱(SELDI- TOF- MS)技术检测卵巢癌患者和健康人血清样本的蛋白质谱,同时采用人工神经网络筛选差异蛋白以建立诊断模型。结果　用SELDI -TOF -MS技术和H4蛋白芯片从47例卵巢癌和29名健康人血清中,筛选出4个有明显表达差异的蛋白,其质荷比(m/z)分别为5 881、7 553、6 652和9 391。用其中的18名健康人和29例卵巢癌患者样本作训练集和交叉验证后,再用筛选出的4个差异蛋白质建立人工神经网络预测模型。然后,对11名健康人和18例卵巢癌患者样本进行盲法测试,以验证该模型。结果显示,我们建立的诊断模型对卵巢癌检测的敏感性为100%,特异性为90 .9%,阳性率为94. 7%。结论　血清蛋白质谱与人工神经网络模型对小样本的卵巢癌诊断具有较高的敏感性和特异性,可扩大样本进行深入的应用性研究。Objective To develop an artificial neural networks tool and use it to identify proteomic patterns in serum for distinguishing ovarian cancer from healthy individuals.Methods 47 ovarian cancer patients and 29 healthy individuals who were randomly divided into training, cross-validation set ( n = 47) and blinding test ( n = 29) set, were analysis by surface-enhanced laser desorption/ ionization (SELDI) mass spectrometry. The peaks were detected and filtrate by Ciphergen proteinchip software 3.0. Using the artificial neural network (ANN) tool and training with the training and cross-validation set, a diagnostic system have been developed for separating ovarian cancer from the healthy groups. The diagnostic system was then used to identify the blinding test set (29 samples).Results Four potential biomarkers were found (5 881、7 553、6 652、9 391m/z), and the diagnostic system separated the ovarian cancer from the healthy samples, with a sensitivity of 100% and a specificity of 90.9%.Conclusion Combination of SELDI with the artificial neural network can get a high sensitivity and specificity approach to identify the ovarian cancer from the healthy samples.

关 键 词：血清蛋白质谱 神经网络模型 应用性研究 SELDI-TOF-MS技术 人工神经网络 诊断模型 飞行时间质谱 激光解吸电离 蛋白芯片 患者样本 卵巢癌患者 血清样本 技术检测 表面增强 差异蛋白 表达差异 预测模型 敏感性 特异性 筛选 

分 类 号：R737.31[医药卫生—肿瘤]