作　　者： ; ;

机构地区： 韩山师范学院化学系

出　　处： 《化学研究与应用》 2006年第4期365-367,共3页

摘　　要： 将lvq神经网络(Learn ing Vector Quantization Neural Networks)用于胃癌组织样品的分类识别,根据胃癌组织及相应正常组织的FTIR光谱的主要特征吸收峰值(包括vas(CH3)、vs(CH2)、δ(CH2)、v(C-O)、vs(PO2-)、vas(PO2-)和vs(核酸,细胞蛋白及膜脂))全部或部分作为网络输入向量,对未知的胃组织样品进行分类识别,结果显示:i)以上述全部七个谱峰为输入向量时,网络经训练学习后,其平均识别正确率最高(达89.3%),表明该网络对胃癌组织样品的分类识别是满意的,完全可作为临床医学的辅助诊断手段;ii)总体上,当作为输入向量的FTIR特征谱峰越多时,则网络的平均分类识别正确率越高;iii)作为输入的FTIR特征谱峰不同时,则网络的平均分类识别正确率也不同。 The learning vector quantization（LVQ）neural network was applied in the recognition of cancerous stomach tissues. The characteristic FFIR peak frequencies（ includingov^as（CH3）、v^s（CH2）、δ（CH2）、v^as（PO2^-）、v（C-O）、v^a（PO2^-） and V^s （ nucleic acid（DNA,RNA） ,cell proteins and membrance lipids） from corresponding stomach cancer tissues were used as the input vectors of the LVQ neural network. The experimental results were given as follows ; i） When the whole FTIR characteristic frequencies mentioned above were all employed as the input vectors, the mean accurate rate of recognition of the LVQ neural network was the best and resached to 89.3% ,which indicated that the LVQ neural network was very satisfactory in the cancer recognition of stomach tissues and could be applied as an auxiliary medical diagnostic method;ii）The LVQ neural network would show higher mean accurate rate of recognition,in general, as more FTIR characteristic frequencies were emdployed as input vectors;iii）The mean accurate rate of recognition of the LVQ neural network would be different, as the input vectors comtained different kind of FTIR characteristic frequencies.

关 键 词： 人工神经网络 学习向量量化 胃癌 傅里叶变换红外光谱 识别

领　　域： [理学] [理学]