可见光光谱用于表寒里热证舌诊的快速分类  

Fast classification of exterior cold and interior heat syndrome,s tongue diagnosis based on spectrometry

作　　者：林凌[1] 李哲[1] 张晶[1] 赵静[1] 熊慧[1,2] 李刚[1] 贾萍[3] 孙景瑞[3] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072 [2]天津工业大学电气工程与自动化学院,天津300160 [3]天津师范大学物理与电子信息学院,天津300387

出　　处：《中华中医药杂志》2012年第1期166-169,共4页China Journal of Traditional Chinese Medicine and Pharmacy

基　　金：国家自然科学基金(No.30973964);天津市科技计划项目(No.10ZXCXSY10400);天津市应用基础及前沿技术研究计划(No.11JCZDJC17100)~~

摘　　要：目的:提出一种应用光谱法来反映舌部信息的方法,最终达到将中医证型客观化的目的。方法:使用光谱仪采集舌部的光谱数据,再通过偏最小二乘和支持向量机的方法,找出光谱数据与中医证型之间的联系,从而实现中医证型客观化。结果:实验用可见光光谱,对采集到的53例表寒里热证患者、37例健康人作分析,分别使用支持向量机和偏最小二乘的方法对数据进行分类和预测,两种方法都可以100%正确地将两类数据区分开,但从预测精度上来讲,支持向量机的预测误差平方均值为0.07734,要优于偏最小二乘的0.13539,所以支持向量机的方法用于可见光光谱的分类要更优一些。结论:提出了中医舌诊的客观化的一种全新研究思路,并为进行中医症候的分类打下基础。Objective： This paper puts forward a method to observe the information the tongue carrying using spectrometry,finally achieve the goal of externalization of application of syndrome differentiation.Methods： This paper collected spectrum data by instrumental means,and seek for association between spectrum data and syndrome differentiation,so that achieve externalization of application of syndrome differentiation.Results： In the experiment,we used visible spectra and collected 53 exterior cold and interior heat patients,37 healthy samples.For data modeling,used partial least square（PLS） and support vector machine（SVM）,both of the method can predict the two sorts of data perfectly with 100%,but considering mean squared error（MSE）,MSE of SVM was 0.07734 which was better than PLS＇s 0.13539.So SVM is a better way to modeling spectrum data.Conclusion： This paper provides a whole new research method of the tongue objectivation of traditional Chinese medicine,and lay the foundation for the classification of traditional Chinese medicine symptoms.

关 键 词：光谱学 舌诊 偏最小二乘法 支持向量机 中医症候 

分 类 号：R241[医药卫生—中医诊断学]