脉搏图像化检测方法对精神疲劳状态的识别
|
摘要
疲劳是指机体在一定环境条件下,由于过于长时间或过度紧张的体力或脑力劳动引起的劳动效率趋向下降的状态。精神疲劳是多种病症的起源。人的大脑长期和过重的超负荷运行,长期处于疲劳状态下,会降低或损害生理功能,引发各种疾病,导致免疫机能的下降、内分泌失调等,这样感冒、心血管病、糖尿病等免疫系统疾病就会趁虚而入。因此,防止疲劳是杜绝各种疾病入侵的关键。由于疲劳状态通常并不伴有明显的病理表现,因而在临床医学检查中难以发现和评判。
脉搏信号包含着丰富的人体生理信息,它反映了身体各子系统的生理状态和病理变化。通过对脉搏触觉信息的分析处理有望获取反映人体健康状态的特征指标,实现对精神疲劳的诊断与程度分级。该研究对于中医脉诊客观化、疾病预防控制、家庭保健,以及避免疲劳驾驶防止事故发生均具有重要而积极的意义。
本文在综述国内外研究现状的基础上,系统地分析了中医脉象研究的工作流程和工作方法。采用脉搏图像化检测装置采集脉搏图像数据。在设计的精神疲劳实验中,组织了30位受试者做三位数组的加减运算,在无人干扰的环境下持续做三个小时的运算,以期受试者能够达到精神疲劳的状态,然后采集志愿者精神疲劳前后的脉搏动态图像并对脉搏视频图像进行相关分析从而提取脉搏信号,然后对脉搏信号进行低通滤波,实现脉搏信号的消噪和去基线漂移。并且根据脉象信号的产生机理、性质,提取出其中与人体生理和病理变化密切相关的特征参量,其中包括功率谱峰值、功率谱峰值频率、功率谱重心、功率谱重心频率、谱能比、倒谱零分量、第一倒谱谐波的幅值与倒谱零分量之比、0-20Hz以内脉搏信号的四组小波包分解能量特征向量。其次应用现代谱估计方法中的参数谱估计对两类脉搏信号的AR参数谱最大值对精神疲劳前后状态进行识别,分类正确率达到90%。
经对60个样本(精神疲劳前后各30组)的LDA分类识别检验,平均正确率达到81.7%。采用BP神经网络方法对神疲劳前后两种脉象进行识别的确率平均达到95%。结果表明:通过脉搏图像化检测装置采集得到的脉搏图像信号经过算法处理得出脉搏信号,运用智能信号处理和现代信号处理方法进行特征提取,最后利用线性判别式分析和BP神经网络进行模式分类,对精神疲劳状态的识别达到了较好的效果,最后对本论文所研究的课题做了结论和展望。
Fatigue refers to an unhealthy state of human body under certain environment conditions which caused by too much prolonged or excessive physical or mental work and which can lead to decline of working efficiency. Mental fatigue is the origin of many diseases. Human brain under the excessive and overloaded tasks especially meatal tasks for a long period can cause the decline or damage of physical function.The state of chronic fatigue will lead to the decrease of immune system and endocrine disorders and so on. In that case, variety of diseases would be infected, such as influenza, cardiovascular disease, diabetes and other diseases. Therefore, avoiding the fatigue is the key to prevent the invasion of various diseases. Since fatigue is usually not accompanied with clear pathological features. Therefore, it's difficult to detect the fatigue state through clinical examination and evaluation.
Pulse signal contains abundant human physiological information, which reflects the various physical and physiological changes in subsystems of hunman body. Through pulse feeling analysis can get the information which reflects the state of human health and can be applied on fatigue state judging and classifation the level of mental fatigue. The research on objectivizing of traditional Chinese pulse diagnosis have significant value on disease prevention, family health, accidents prevention and fatigue prevention and so on.
This article summarized the present research situation both in domestic and abroad, then the research process and method on Chinese pulse checkings are systematically analyzed. Pulse image detecting device which was developed by our research team is applied to acquire the pulse image data of testees. In mental fatigue test, 30 volunters are organized to join the test designed as three arrays addition and subtraction operations for 3 hours in a quiet environment, in that case it was supposed that 30 volunters are all in mental fatigue situation. The pulse image data which acquired as vedio areprocessed with image correlation algorithm and the pulse signals of each volunter is obtained, pulse signals are processed by low pass filter for de-noising and non-drifting. According to the mechanism and characteristics of pulse signal, several features wich can reflect the physical and pathological relationship between human body and pulse signal, including the power spectrum peak value, peak frequency, center of gravity, gravity frequency of power spectrum, the value of SER, zero weight of cepstral, the ratio of cepetral first harmorous wave amplitude and zero weight of cepstral, energy value of four frequency bands form OHz to 20Hz which decomposed by wavelet-packet. Mental fatigue identification is also applied the morden signal processing method, through applying the maxim value of AR parameter spectrum, the accurate rate of the two patterns can reach 90%.
After the classification test for 60 groups of samples which belong to 30 volunters by applying LDA examination, the average accurate rate reach to 81.7%.Though BP neural network classification for two patterns, the accurate rate reach to95%. Through these two kinds of classification methods, it is proved that our test and signal processing methods are efficient and objective. Finally the conclusion and the forecast are drawn for our subject.
脉搏信号包含着丰富的人体生理信息,它反映了身体各子系统的生理状态和病理变化。通过对脉搏触觉信息的分析处理有望获取反映人体健康状态的特征指标,实现对精神疲劳的诊断与程度分级。该研究对于中医脉诊客观化、疾病预防控制、家庭保健,以及避免疲劳驾驶防止事故发生均具有重要而积极的意义。
本文在综述国内外研究现状的基础上,系统地分析了中医脉象研究的工作流程和工作方法。采用脉搏图像化检测装置采集脉搏图像数据。在设计的精神疲劳实验中,组织了30位受试者做三位数组的加减运算,在无人干扰的环境下持续做三个小时的运算,以期受试者能够达到精神疲劳的状态,然后采集志愿者精神疲劳前后的脉搏动态图像并对脉搏视频图像进行相关分析从而提取脉搏信号,然后对脉搏信号进行低通滤波,实现脉搏信号的消噪和去基线漂移。并且根据脉象信号的产生机理、性质,提取出其中与人体生理和病理变化密切相关的特征参量,其中包括功率谱峰值、功率谱峰值频率、功率谱重心、功率谱重心频率、谱能比、倒谱零分量、第一倒谱谐波的幅值与倒谱零分量之比、0-20Hz以内脉搏信号的四组小波包分解能量特征向量。其次应用现代谱估计方法中的参数谱估计对两类脉搏信号的AR参数谱最大值对精神疲劳前后状态进行识别,分类正确率达到90%。
经对60个样本(精神疲劳前后各30组)的LDA分类识别检验,平均正确率达到81.7%。采用BP神经网络方法对神疲劳前后两种脉象进行识别的确率平均达到95%。结果表明:通过脉搏图像化检测装置采集得到的脉搏图像信号经过算法处理得出脉搏信号,运用智能信号处理和现代信号处理方法进行特征提取,最后利用线性判别式分析和BP神经网络进行模式分类,对精神疲劳状态的识别达到了较好的效果,最后对本论文所研究的课题做了结论和展望。
Fatigue refers to an unhealthy state of human body under certain environment conditions which caused by too much prolonged or excessive physical or mental work and which can lead to decline of working efficiency. Mental fatigue is the origin of many diseases. Human brain under the excessive and overloaded tasks especially meatal tasks for a long period can cause the decline or damage of physical function.The state of chronic fatigue will lead to the decrease of immune system and endocrine disorders and so on. In that case, variety of diseases would be infected, such as influenza, cardiovascular disease, diabetes and other diseases. Therefore, avoiding the fatigue is the key to prevent the invasion of various diseases. Since fatigue is usually not accompanied with clear pathological features. Therefore, it's difficult to detect the fatigue state through clinical examination and evaluation.
Pulse signal contains abundant human physiological information, which reflects the various physical and physiological changes in subsystems of hunman body. Through pulse feeling analysis can get the information which reflects the state of human health and can be applied on fatigue state judging and classifation the level of mental fatigue. The research on objectivizing of traditional Chinese pulse diagnosis have significant value on disease prevention, family health, accidents prevention and fatigue prevention and so on.
This article summarized the present research situation both in domestic and abroad, then the research process and method on Chinese pulse checkings are systematically analyzed. Pulse image detecting device which was developed by our research team is applied to acquire the pulse image data of testees. In mental fatigue test, 30 volunters are organized to join the test designed as three arrays addition and subtraction operations for 3 hours in a quiet environment, in that case it was supposed that 30 volunters are all in mental fatigue situation. The pulse image data which acquired as vedio areprocessed with image correlation algorithm and the pulse signals of each volunter is obtained, pulse signals are processed by low pass filter for de-noising and non-drifting. According to the mechanism and characteristics of pulse signal, several features wich can reflect the physical and pathological relationship between human body and pulse signal, including the power spectrum peak value, peak frequency, center of gravity, gravity frequency of power spectrum, the value of SER, zero weight of cepstral, the ratio of cepetral first harmorous wave amplitude and zero weight of cepstral, energy value of four frequency bands form OHz to 20Hz which decomposed by wavelet-packet. Mental fatigue identification is also applied the morden signal processing method, through applying the maxim value of AR parameter spectrum, the accurate rate of the two patterns can reach 90%.
After the classification test for 60 groups of samples which belong to 30 volunters by applying LDA examination, the average accurate rate reach to 81.7%.Though BP neural network classification for two patterns, the accurate rate reach to95%. Through these two kinds of classification methods, it is proved that our test and signal processing methods are efficient and objective. Finally the conclusion and the forecast are drawn for our subject.
引文
[1]黄世林,孙明异.中医脉象研究[M].北京:人民卫生出版社,1986.
[2]费兆馥.现代中医脉诊学.北京:人民卫生出版社,2003.
[3]蔡轶珩,沈兰荪,黄祥林.脉象分析仪的研究进展.电子测量与仪器学报,2002,16(4):23-27.
[4]费兆馥,张志枫.中医脉诊的图像化和定量化.自然杂志,1995,17(5):269-274.
[5]沈凤麟,陈和.生物医学随机信号处理.中国科学技术大学出版社.
[6]王宏禹.非平稳随机信号分析与处理.国防工业出版社,1999.
[7]罗山鹰,娄文忠.一种用于人体脉搏的信号采集系统.兵工自动化,2001,20(2):19-24.
[8]J.Z.Wu,R.G.Dong,S.Rakheja,A.W.Schopper,W.P.Smutz.A Structural Fingertip Model for Simulating of the Biomechanics of Tactile Sensation.Medical Engineering & Physics.26(2004),165-175.
[9]张东青,王元昔.新型自检脉搏传感器的研制.传感器技术,2001,12(5):47-52.
[10]Seppo Nissila,Mika Sorvisto,Hannu Sorvoja.Non-invasive Pulse Pressure Measurement Based on the Electronic Palpation Method.Proceedings of the 20th Annual International Conference of the IEEE Engineering in Medicine and Biology Society,1998,20(4):104-111.
[11]Young-Zoon Yoon,Myeong-Hwa Lee,Kwang-Sup Soh.Pulse type classification by varying contact pressure.IEEE Engineering in Medicine and Biology,2000,6(9):106-110.
[12]刘克俭,谢红兵,毛福英,蔡荣泰.轮船驾驶作业人员疲劳特征的研究.水利电力劳动保护,1997,3(1):31-36.
[13]杨福生,高上凯.生物医学信号处理.北京:高等教育出版社,1989.
[14]章毓晋.图像视觉与计算机视觉.北京:清华大学出版社,2000.8.
[15]龚沛曾,陆慰民,杨志强.Visual Basic程序设计教程.第1版.北京:高等教育出版社,2000.
[16]徐飞,施晓红等MATLAB应用图像处理西安电子科技大学出版社,2002.5
[17]Lygouras,J.N,Tsalides,P.G.Optical-fiber finger photo-plethysmograph using digital techniques.IEEE Sensor Journal,2002,2(1):20-25.
[18]罗志昌,张松,杨文鸣.脉搏波波性特征的研究.西北工业大学学报, 1996,22(1):71-79.2001,31(1):21-25.
[19]Ling Y.& Peter C.,Frequency distribution of human pulse spectra,IEEE Trans BME,1985,32(3):245-246.
[20]胡广书.数字信号处理 理论,算法与实现.清华大学出版社,1997.
[21]李勇,徐震.辅助现代工程数字信号处理.第1版,西安:西安电子科技大学出版社,2002.
[22]张镜人,杨天权,郑秀春,孙瑾.脉象图频域指标和时域指标的典型相关分析.辽宁中医杂志,1995,22(9):393-395.
[23]王炳和,罗建,相敬林.人体脉搏功率谱分析与中医脉诊机理研究,
[24]张世枢,洪志平.中医脉象分类的频谱特征研究.辽宁中医杂志,1997,21(2):24.
[25]Irina Hlimonenko,Kalju Meigas,Rein Vahisalu,Waveform Analysis of Peripheral Pulse Wave Detected in the Fingertip with Photo-plethysmograph,Measurement in Biomedicine,2003,3(2):49-52.
[26]王炳和,张效民,相敬林,职利琴.倒谱与解卷积技术在人体脉搏系统分析中的应用.西北工业大学学报,2000,18(4):600-603.
[27]王炳和,罗建,相敬林等.人体脉搏功率谱分析与中医脉诊机理研究.西北大学学报(自然科学版),2001,2(31):21-25.
[28]张爱华,杨凤霞,王润东.基于脉象信号的亚健康状态识别.兰州理工大学学报(自然科学版),2006,32(6):82-84.
[29]宋建勤,董彦武,吴胜举.倒谱技术在人体脉搏信号分析中的应用.陕西师范大学学报(自然科学版).1997,25(2):38-42.
[30]王炳和,人体脉搏系统建模与脉搏信息处理方法的研究.[西北工业大学博士学位论文],西安:西北工业大学,1999.24-27.
[31]王炳和,相敬林.基于AR模型的人体脉象信号模糊聚类研究.应用声学,2001,20(5):21-25.
[32]胡昌华,周涛,夏启兵等基于的系统分析与设计一时频分析.第四版.西安:西安电子科技大学出版社,2002,18-51.
[33]蔡坤宝,许继勇.中医脉象信号的参数化双谱估计及切片分析.重庆工学院学报(自然科学版)2007,21(3):75-79.
[34]Giannakis.G.B.On the identifiability of non-Gaussian ARMA models using cumulants.IEEETrans.Automatic Contr,1990,35(1):18-26.
[35]Swami.A,Mende]J.M.Identifiability of AR parameters of an ARMA process using cumulants.IEEE Trans.Automatic Control.1992,37(2):68-73.
[36]李玉韩,蔡坤宝,张继红.双谱估计在中医脉象信号分析中的应用.重庆大学学报(自然科学版)2006.5,129(5):142-147.
[37]秦前清.实用小波分析.西安电子科技大学出版社,1998.
[38]胡昌华,张军波,夏伟等.基于的系统分析与设计一小波分析,第1版,西安:西安电子科技大学出版社,2001.
[39]胡昌华,张军波等.基于的系统分析与设计-小波分析.西安西安电子科技大学出版社.
[40]Matlab6.5辅助小波分析与应用 电子工业出版社.
[41]谢家宇,蔡坤宝,王永东.连续小波变换在中医脉象信号处理中的应用.重庆大学学报,2003,26(1):66-68.
[42]周霞,蔡坤宝.中医脉象信号的短时傅里叶分析.重庆大学学报,2003,26(10):47-51.
[43]柴新禹,吴朝霞.时频分析方法及其在医学信号处理中的应用.生物医学工程学杂志,2001,18(1):138-144.
[44]张丽琼,王炳和.基于小波变换的脉象信号特征提取方法.数据采集与处理.2004,19(3):126-131
[45]边肇祺,张学工等.模式识别.北京:清华大学出版社,2000.
[46]王爱民,张维廉.脉象计算机识别和分类研究初探.生物医学工程学杂志,1992,9(1):86-90.
[47]徐方维,蔡坤宝.人工神经网络在中医脉象信号检测中的应用.重庆大学学报(自然科版),2004,27(8):35-39.
[48]胡家宁,阎述池等,脉象特征人工神经网络分类器,中国医科大学学报,1996,25(6):571-575.
[49]岳沛平,李训铭.基于小波分析和BP神经网络识别的中医脉象信号辨识系统计算机与现代化.2005,No.2:38-42.
[50]Shinichi Horikawa,Takeshi Furuhashi,Oshiki Uchikawa.On fuzzy modeling using fuzzy neural networks with the back-propagation algorithm.IEEE Trans.On Neural Networks,1992,3(5):796-801.
[51]楼顺天,施阳.基于MATLAB的系统分析与设计——神经网络.西安:西安电子科技大学出版社.1998.
[52]程自峰,伟哲.人工神经网络在医学信号检测与分析中的应用.医疗装备,1996,15(3):188-192.
[53]胡家宁,阎述池等,脉象人工神经网络分析系统模型,中国医科大学学报,1997,26(2):134-137.
[2]费兆馥.现代中医脉诊学.北京:人民卫生出版社,2003.
[3]蔡轶珩,沈兰荪,黄祥林.脉象分析仪的研究进展.电子测量与仪器学报,2002,16(4):23-27.
[4]费兆馥,张志枫.中医脉诊的图像化和定量化.自然杂志,1995,17(5):269-274.
[5]沈凤麟,陈和.生物医学随机信号处理.中国科学技术大学出版社.
[6]王宏禹.非平稳随机信号分析与处理.国防工业出版社,1999.
[7]罗山鹰,娄文忠.一种用于人体脉搏的信号采集系统.兵工自动化,2001,20(2):19-24.
[8]J.Z.Wu,R.G.Dong,S.Rakheja,A.W.Schopper,W.P.Smutz.A Structural Fingertip Model for Simulating of the Biomechanics of Tactile Sensation.Medical Engineering & Physics.26(2004),165-175.
[9]张东青,王元昔.新型自检脉搏传感器的研制.传感器技术,2001,12(5):47-52.
[10]Seppo Nissila,Mika Sorvisto,Hannu Sorvoja.Non-invasive Pulse Pressure Measurement Based on the Electronic Palpation Method.Proceedings of the 20th Annual International Conference of the IEEE Engineering in Medicine and Biology Society,1998,20(4):104-111.
[11]Young-Zoon Yoon,Myeong-Hwa Lee,Kwang-Sup Soh.Pulse type classification by varying contact pressure.IEEE Engineering in Medicine and Biology,2000,6(9):106-110.
[12]刘克俭,谢红兵,毛福英,蔡荣泰.轮船驾驶作业人员疲劳特征的研究.水利电力劳动保护,1997,3(1):31-36.
[13]杨福生,高上凯.生物医学信号处理.北京:高等教育出版社,1989.
[14]章毓晋.图像视觉与计算机视觉.北京:清华大学出版社,2000.8.
[15]龚沛曾,陆慰民,杨志强.Visual Basic程序设计教程.第1版.北京:高等教育出版社,2000.
[16]徐飞,施晓红等MATLAB应用图像处理西安电子科技大学出版社,2002.5
[17]Lygouras,J.N,Tsalides,P.G.Optical-fiber finger photo-plethysmograph using digital techniques.IEEE Sensor Journal,2002,2(1):20-25.
[18]罗志昌,张松,杨文鸣.脉搏波波性特征的研究.西北工业大学学报, 1996,22(1):71-79.2001,31(1):21-25.
[19]Ling Y.& Peter C.,Frequency distribution of human pulse spectra,IEEE Trans BME,1985,32(3):245-246.
[20]胡广书.数字信号处理 理论,算法与实现.清华大学出版社,1997.
[21]李勇,徐震.辅助现代工程数字信号处理.第1版,西安:西安电子科技大学出版社,2002.
[22]张镜人,杨天权,郑秀春,孙瑾.脉象图频域指标和时域指标的典型相关分析.辽宁中医杂志,1995,22(9):393-395.
[23]王炳和,罗建,相敬林.人体脉搏功率谱分析与中医脉诊机理研究,
[24]张世枢,洪志平.中医脉象分类的频谱特征研究.辽宁中医杂志,1997,21(2):24.
[25]Irina Hlimonenko,Kalju Meigas,Rein Vahisalu,Waveform Analysis of Peripheral Pulse Wave Detected in the Fingertip with Photo-plethysmograph,Measurement in Biomedicine,2003,3(2):49-52.
[26]王炳和,张效民,相敬林,职利琴.倒谱与解卷积技术在人体脉搏系统分析中的应用.西北工业大学学报,2000,18(4):600-603.
[27]王炳和,罗建,相敬林等.人体脉搏功率谱分析与中医脉诊机理研究.西北大学学报(自然科学版),2001,2(31):21-25.
[28]张爱华,杨凤霞,王润东.基于脉象信号的亚健康状态识别.兰州理工大学学报(自然科学版),2006,32(6):82-84.
[29]宋建勤,董彦武,吴胜举.倒谱技术在人体脉搏信号分析中的应用.陕西师范大学学报(自然科学版).1997,25(2):38-42.
[30]王炳和,人体脉搏系统建模与脉搏信息处理方法的研究.[西北工业大学博士学位论文],西安:西北工业大学,1999.24-27.
[31]王炳和,相敬林.基于AR模型的人体脉象信号模糊聚类研究.应用声学,2001,20(5):21-25.
[32]胡昌华,周涛,夏启兵等基于的系统分析与设计一时频分析.第四版.西安:西安电子科技大学出版社,2002,18-51.
[33]蔡坤宝,许继勇.中医脉象信号的参数化双谱估计及切片分析.重庆工学院学报(自然科学版)2007,21(3):75-79.
[34]Giannakis.G.B.On the identifiability of non-Gaussian ARMA models using cumulants.IEEETrans.Automatic Contr,1990,35(1):18-26.
[35]Swami.A,Mende]J.M.Identifiability of AR parameters of an ARMA process using cumulants.IEEE Trans.Automatic Control.1992,37(2):68-73.
[36]李玉韩,蔡坤宝,张继红.双谱估计在中医脉象信号分析中的应用.重庆大学学报(自然科学版)2006.5,129(5):142-147.
[37]秦前清.实用小波分析.西安电子科技大学出版社,1998.
[38]胡昌华,张军波,夏伟等.基于的系统分析与设计一小波分析,第1版,西安:西安电子科技大学出版社,2001.
[39]胡昌华,张军波等.基于的系统分析与设计-小波分析.西安西安电子科技大学出版社.
[40]Matlab6.5辅助小波分析与应用 电子工业出版社.
[41]谢家宇,蔡坤宝,王永东.连续小波变换在中医脉象信号处理中的应用.重庆大学学报,2003,26(1):66-68.
[42]周霞,蔡坤宝.中医脉象信号的短时傅里叶分析.重庆大学学报,2003,26(10):47-51.
[43]柴新禹,吴朝霞.时频分析方法及其在医学信号处理中的应用.生物医学工程学杂志,2001,18(1):138-144.
[44]张丽琼,王炳和.基于小波变换的脉象信号特征提取方法.数据采集与处理.2004,19(3):126-131
[45]边肇祺,张学工等.模式识别.北京:清华大学出版社,2000.
[46]王爱民,张维廉.脉象计算机识别和分类研究初探.生物医学工程学杂志,1992,9(1):86-90.
[47]徐方维,蔡坤宝.人工神经网络在中医脉象信号检测中的应用.重庆大学学报(自然科版),2004,27(8):35-39.
[48]胡家宁,阎述池等,脉象特征人工神经网络分类器,中国医科大学学报,1996,25(6):571-575.
[49]岳沛平,李训铭.基于小波分析和BP神经网络识别的中医脉象信号辨识系统计算机与现代化.2005,No.2:38-42.
[50]Shinichi Horikawa,Takeshi Furuhashi,Oshiki Uchikawa.On fuzzy modeling using fuzzy neural networks with the back-propagation algorithm.IEEE Trans.On Neural Networks,1992,3(5):796-801.
[51]楼顺天,施阳.基于MATLAB的系统分析与设计——神经网络.西安:西安电子科技大学出版社.1998.
[52]程自峰,伟哲.人工神经网络在医学信号检测与分析中的应用.医疗装备,1996,15(3):188-192.
[53]胡家宁,阎述池等,脉象人工神经网络分析系统模型,中国医科大学学报,1997,26(2):134-137.