【作者】 张香焕；

【导师】 黄岳山；

【作者基本信息】 华南理工大学 ， 生物医学工程， 2011， 硕士

【摘要】 当今的社会,“亚健康”已严重影响人们的生活、工作和健康。研究和对亚健康进行干预,已成为人类医学的重要课题,对保护人类健康意义重大。国内外对亚健康的判断主要还是通过问卷调查等的定性评价方法。人们希望结合一些常规生理检测实现一定程度的定量评价,中医的脉象蕴涵着丰富的生理病理信息,适合用于亚健康评价,已形成研究热点。本文通过提取脉搏信号中的有用信息,评估和诊断个体的身心状态,为亚健康的诊断,提供一些客观依据。本文的主要工作有:综述了亚健康的形成原因、产生的危害及其评测方法、中医脉象的形成机理及特征、临床意义,并从中医脉诊角度分析了利用脉搏信号中蕴含的信息来识别亚健康状态的可行性。研究了信号的预处理方法。利用具有良好时间-尺度的小波变换法,经验模态分解法,中值滤波法及整系数法对信号做预处理。同时从处理后的波形图,信噪比和处理所需时间三个方面进行算法优越性的对比。整系数滤波法处理时间较短,适用于实时处理;小波阈值滤波具有较高的信噪比;经验模态分解法相对整系数滤波法,处理时间优势不明显,但EMD可以有效的避免整系数去噪带来的边界效应。运用小波多分辨率的特性和经验模态分解,结合实际经验,提取脉图面积K值,频谱峰值,谱能比特征量。运用模式识别中的支持向量机方法,对30个样本进行分类处理。其中脉图面积K值的识别率为80%,频谱峰值的识别率为60%,谱能比的识别率为70%。结果表明,在一定程度上,脉搏信号的特征可以作为亚健康的判据。更多还原

【Abstract】 In modern society ,“sub-health”has seriously affect people’s life, work and health. Studying and intervening of”sub-health”is an important subject of human medicine and of great significance to protect human health .The judgment of sub-health at home and abroad is qualitative evaluation based on chemical, biological and questionnaire method. People hope to quantitative evaluation in a certain degree with some conventional physical testing . In Chinese Medicine, pulse wave contains much wealth of human physiology and pathology information, that can be useful for sub-health evaluation, and formed research focus. This paper aims to extract useful information in the pulse wave, that is used to assessment and diagnosis of individual physical , psychological condition and provide some objective basis for sub-health diagnosis. The main work:Reviewed the sub-health from the formation reasons, dangers and evaluation methods; the formation mechanism of pulse, waveform characteristics, clinical significance. Analyzed the relationship between sub-health and pulse signal from Chinese Pulse Diagnosis based on the information of pulse wave .Studying of the signal preprocessing method. Using wavelet transform, empirical mode decomposition, median filtering method and the whole coefficient filter method do preprocessing. Meanwhile, compared the algorithm from SNR, processing time and waveform. The the whole coefficient filter method has short processing time ,that can be using in real-time processing. Wavelet threshold method has high SNR. Empirical mode decomposition can effectively avoid the boundary effect compared with the the whole coefficient filter method.Using wavelet and EMD, combined with practical experience extract the pulse image area of K, the main spectrum peaks, SER. SVM is used to identify sub-health from the information of 30 persons containing sub-health and health. The recognition accuracy is up to 80% by using the K characteristics and 60% by using the main spectrum peaks, and 70% by using the SER. The results show that, pulse signal feature can be used as the criterion of sub-health.更多还原