毛细管电泳-小波神经网络法辅佐诊断乳腺癌的研究  

Application of the Combination of Capillary Electrophoresis with Wavelet Neural Network in the Assisted Clinical Diagnosis of Breast Cancer

作　　者：熊建辉[1] 郑育芳[1] 张普敦[1] 石先哲[1] 杨军[1] 张玉奎[1] 许国旺[1] 

机构地区：[1]中国科学院大连化学物理研究所国家色谱研究分析中心,大连116011

出　　处：《高等学校化学学报》2003年第5期803-807,共5页Chemical Journal of Chinese Universities

基　　金：国家自然科学基金(批准号:29775024);大连市科学基金;中国科学院知识创新工程领域前沿项目资助

摘　　要：采用毛细管电泳法测定了46个健康人和26个乳腺癌病人尿样中的13种正常核苷和修饰核苷,以小波神经网络作为模式识别工具对健康人和乳腺癌病人的分类作了研究,随机选取的训练集的识别率达到100％,相应的预测集判别率正确性在96％以上,与经典的前向多层神经网络相比,小波神经网络具有更强的信息提取和逼近能力。研究结果还表明,小波神经网络的预测能力强于主成分分析和线性判别分析,毛细管电泳法与小波神经网络的结合有望成为乳腺癌的辅助诊断手段。Thirteen kinds of normal and modified nucleosides were determined in urine samples from 46 healthy persons and 26 breast cancer patients by capillary electrophoresis. A wavelet neural network model has been used as a powerful pattern recognition tool to distinguish breast cancer patients from healthy persons. The recognition rate for the training set reached to 100% and above 96% of people in the predicting set were correctly classified. Compared with standard backpropagation neural network, wavelet neural network had stronger abilities of information extraction and approximation. The results also demonstrated that the predicting ability of wavelet neural network was higher than those of principal component analysis and linear discriminant analysis. The combination of capillary electrophoresis and wavelet neural network was expected to be an assisted tool for the clinical diagnosis of breast cancer.

关 键 词：毛细管电泳 小波神经网络法 诊断 乳腺癌 核苷 尿 

分 类 号：R737.9[医药卫生—肿瘤] R730.43[医药卫生—临床医学]