摘要
随着人类社会的不断发展进步,人们的生活水平也逐步提高,同时生活节奏也越来越快，促使人们对生命健康质量有了更高的要求，然而长时间从事脑力劳动、睡眠不足、情绪压抑、过度疲劳会引发内脏功能失调、心血管功能紊乱、腺体分泌异常，这些都影响着人们的正常生活。现代医学通过生理信号的变化断对病人身体状进行评估，然而单一信号(心电、脉搏等)对人体的生理状态的变化判别能力较差。本文从两种生理信号同步分析角度出发，通过探讨人体生理信号之间的内在联系，提出了基于心电和脉搏信息融合分析的研究方法。主要内容如下：
（1）通过同步采集平台对心电信号和脉搏信号进行采集，对同步采集的信号做处理，得到疲劳实验前后的心电信号和脉搏信号。
（2）选择合适的滤波器对心电信号和脉搏信号进行预处理，观察时域和频域上基线漂移、工频干扰、肌电干扰的滤除状况，得到准确的波形。
（3）分析实验前后心电的R波，T波以及脉搏的主波和重搏波前波等特征，利用t检验检测各个特征信息变化的显著性水平，其显著性水平p<0.01；通过R波和脉搏主波间期分析心率变异性(HRV)和脉率变异性(PRV)，对HRV和PRV的间期均值、SDDN、甚低频（VLF）、低频（LF）、高频（HF）、相对频谱分布（LF/HF）和相关系数（ρ）等特征在同一实验状态和不同实验状态下分析。利用t检验各个特征发现有统计学意义p<0.05。
最后，通过提取到的心电和脉搏的重要特征，对人体的生理特性进行分析，发现其心率变异性和脉率变异性的相关特性，以及二者和植物性神经之间的关系。并对比不同的信息融合方法，利用特征级融合对心电信号和脉搏信号的视觉疲劳状态进行分析，利用支持向量机(SVM)法对34例样本(正常17例，视觉疲劳17例)的心电和脉搏融合特征进行分类，达到了较高的分类效果。
研究结果表明，心电信号和脉搏信号在人体中有极其紧密的相关特征，在利用心电、脉搏信号反映人体疲劳状态的同时；利用心电和脉搏的相关性对疲劳状态的研究和分析是能够达到更好的效果。
With the ceaseless development of human society, people's living standardsgradually improved and spurred to greater demands on the quality of life and health.However, engaging in mental work for a long time, lack of sleep or emotionaldistress, fatigue and irregular life can cause internal function disorder,cardiovascular disorders and unusual glandular secretion. This could affect thepeople's normal life seriously. Physiological signal is used to determine the healthstatus in the modern medicine. However, the single signal (ECG, pulse, etc.) waspoor to discriminate the body's physiological state changes. Proceeding fromsimultaneous analysis of two kinds of physiological signals and exploring thecorrelation between human physiological signals, this paper presents a fusionanalysis method based on ECG and pulse signals. The main works were shown asfollowing:
(1) The ECG and pulse signals were acquired synchronously before and after thefatigue experiment through the data acquisition system.
(2) Preprocess ECG and pulse signal with the appropriate filter, and eliminate thebaseline drift with the frequency50Hz and EMG interference. Then we can obtainaccurate waveforms.
(3) The characteristics of the R-wave, T wave, the peak value and the tidal wavepeak of the pulse signals were analyzed. The t test was used to evaluate thesignificance level of each characteristic changes and the significance level are allp<0.01. The R-R intervals and P-P intervals could be extracted from the ECG andthe pulse signals. Analysis the change before and after the visual fatigue experimentin the time domain and frequency domain feature extraction. The mean interval ofHRV and PRV characteristics, SDDN, very low frequency (VLF), low frequency(LF), high frequency (HF), the relative spectral distribution (LF/HF) and correlationwere analyzed under the same experimental state and the different experimentalstates respectively, The t test results have demonstrated each feature has statisticalsignificance p<0.01.
Finally, the characteristics of the ECG and pulse were used to analyze thehuman physiological property. The results have shown that there is correlationproperty between HRV/PRV and Autonomic Nerve in VDT Fatigue. Comparingdifferent information fusion methods and using feature fusion to analyze the VDT Fatigue, the accuracy rate of classification reaches up to100%by using SupportVector Machine for the combination features of ECG and pulse wave signals, whichsurpassed the accuracy rate classified by one kind of biomedical signal.
The results have shown that ECG and pulse signal in the human body are veryclosely related characteristics. Not only the ECG and the pulse signals can reflect thefatigue state, but also the correlation between ECG and pulse signals have bettereffect on the research and analysis of the fatigue state.
引文
[1]曹海.英年早逝过劳死[J].现代保健.2004，4(7):5-7
[2]张崇,郑崇勋,裴晓梅等.生理性精神疲劳的多参数脑电功率谱分析[J].生物医学工程学杂志,2009,26(1):162-163.
[3] Jiang hu.JF.Reseach of Drowsiness in Driiving Based on EEG[C].In2010ThirdInternational Symposiun on Electronic.Commerce and Security.2010.328-331
[4] Subasi A. Application of Classical and Model-Based Spectral Methods toDescribe the State of Alertness in EEG [J].Journal of Medical Systems.2005,29(5):437-486
[5] Lin CT, Chen YC, Huang TY, et al. Development of Wireless Brain ComputerInterface with Embedded Multitask Scheduling and its Application onReal-Time Driver’s Drowsiness Detection and Warning[J]. In: IEEETransactions on Biomedical Engineering.2008,1582-1591.
[6]颜松,魏建勤,吴永红.汽车驾驶员瞌睡状态脑电波特征提取的研究[J].中国生物医学工程学报.2005.1（24）:110-114
[7]熊敏,刘雄飞.基于多孔算法的心电图QRS波检测[J].计算机仿真.2011,12:244-248
[8] F. Portet, A.I. Hern indez, G.Carrault. Evaluation of real-time QRSdetectionalgorithms in variable contexts [J].Biol. Eng. Comput.2005，43:379-385
[9]李冉,方滨,沈毅,孙崇正,王普.ECG中P波检测的Prony分析方法[J].生物医学工程杂志.2008,25(6):1271-1275
[10] Metin Akay, J.anderw Daubenspeck. Respiratory Related Evoked Responses toGraduated Pressure Pulses using Wavelet Transform Methods. Annals ofBiomedical Engineering.2000,28:1126–1135.
[11]王炳和,相敬林.基于AR模型的人体脉象信号模糊聚类研究[J].2000，20(5):21-25
[12]王炳和,郭红霞.从脉搏信号中准确提取呼吸和心率信息的新方法[J].陕西师范大学学报.2005,33(1):53-55
[13]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[J].2004,19(3):323-328
[14]温小明.基于小波和小波包的脉搏波处理[J].重庆科技学院学报（自然科学版）.2001,10(5):108-110
[15]张爱华，豆小玺，王龙.脉搏信号功率谱分析对精神疲劳状态的识别[J].中国组织工程研究与临床康复2007,11(1):118-120
[16]高嵩,吕巍,张仲.32例冠心病患者心率变异性(HRV)指标SDNN、rMSSD、PNN50的临床研究远临床探讨[J].2009,47(25):124-125.
[17]唐文樑,魏盟.副交感神经系统与心力衰竭.国际心血管杂志[J].2010,4:213-215
[18] Akihiko Uehara,Chinori Kurata,Tosh-ihiko Sugi,et al. Diabetic cardiacautonomic dysfunction: parasympathetic versus sympathetic [J]. Annals ofNuclear Medicine.1999,13(12):95-100.
[19] Patrizia Muroni,Roberto Crnjar,Iole Tomassini Barbarossa. Emotionalresponses to pleasant and unpleasant oral flavour stimuli [J]. Chemistry andmaterials science.2011,4(3):65-71.
[20] D. Petkovi, ojba i.Adaptive neuro-fuzzy estimation of autonomic nervoussystem parameters effect on heart rate variability [J]. Neural Computer andApplic.2011,5(18):1-6.
[21] C. Heinze,U.Trutschel,T.Schnupp,et al.Operator Fatigue Estimation UsingHeart Rate Measures [C].IFMBE.2010.25(4):930-933
[22]焦昆,李增勇,陈铭等.汽车驾驶员驾驶过程中的心率变异性功率谱分析[J].中国生物医学工程学报.2003,22(6):574-576.
[23]李延军,严洪,杨向林,王政.基于心率变异性的精神疲劳研究[J].中国生物医学工程学报.2010,29(1):1-6
[24赵海勇,邱意弘,胡思钧,朱贻盛.冠心病病人脉搏变异性信号的去趋势波动分析[J].航天医学与医学工程.2007,6:455-457
[25]王业泰.基于脉搏信号散点图分析的VDT视疲劳研究[C].振动与噪声测试峰会论文集.2010,44-47
[26]江玲,邵怿,张琇文,张心怡,雍丽.脉图参数评估抑郁症患者植物神经功能特点的探索性研究.中国中医药杂志[J].2011.04:10-11
[27]中医养生学的基本理论.http://www.360doc.com
[26] B M Pannier, A P Avolio, A Hoeks, et al. Methods and devices for measuringarterial compliance in humans[J]. American Journal of Hypertension,2002,15:743–753
[29] Eduardo G, Michele O, Raquel B. Time-varying spectral analysis forcomparison of HRV and PPG variability during tilt table test.2010,8:2579-2582
[30]吴学奎,任立红,丁永生,吴怡之.面向智能服装的多生理信息融合的情绪判别[J].计算机工程及应用.2009,45(33):218-235
[31]沈永增，胡立芳，冯继妙.多元信息融合在驾驶疲劳检测中的应用[J].计算机应用与软件.2012，29（2）：272-274
[32]付华，王雨虹.基于数据挖掘的瓦斯灾害信息融合模型的研究[J].传感器与微系统.2008，27（1）：52-54
[33]杜宁.基于信息融合的新的人脸检测算法[J].电子技术.2011，9
[34]王树亮，阮怀林，张兴良.目标多特征信息的模糊数据关联算法[J].火力与指挥控制.2011，36（10）：127-130
[35]吴蔚，朱家瑞.信息融合技术的一个热点：医学图像融合[J].国外医学放射医学核医学分册.1998，22（3）：103-105
[36]谢杨.原发性传导束退化.世界今日医学杂志[J].2003,3:44-45
[37]赵治月.基于心电脉搏信号的视觉疲劳状态识别方法研究[D].[兰州理工大学硕士学位论文].兰州：兰州理工大学,2009,6:25-26
[38]季忠,秦树人.微弱生物医学信号特征提取的原理与实现[M].第一版.北京:科学出版社.2007
[39]朱洪俊.心电信号零相位滤波.北京生物医学工程[J].2003,4:152-153
[40]袁丽华.陷波滤波器的设计及其应用.自动化与仪器仪表[M].2004.4
[41]董红生.基于平稳小波变换的自适应心电信号去噪方法.中国医疗器械杂志[J].2009.3
[42] Alty, S.R.Angarita-Jaimes, N. Millasseau, S.C, et al. Predicting arterial stiffnessfrom the digital volume pulse wave form [J]. IEEE Transactions on BiomedicalEngineering,2007,54(12):2268-2275.
[43]胡广书.现代信号处理教程[M].北京：清华大学出版社.2004:239-245
[44]吴琳娜,刘少强,汪立林.新型脉搏波检测时域处理方法与系统实现[J].传感器与微系统,2009,27(9):72-74
[45] Zhang Aihua, Chai Long, Dong Hongsheng. QRS Complex Detection of ECGSignal by Using Teager Energy Operator[C]. In: The2nd InternationalConference on Bioinformatics and Biomedical Engineering, Shanghai, China,May.2008:2095-2098
[46] P.Tchofo Dinda, J. Atangana, A. Kamagate1, et al.Effective method ofcharacterization of the phase and intensity profiles of asymmetrically distortedlight pulses in optical fiber systems.The European Physical Journal SpecialTopics,2009,173:121-138
[47]杨华.基于心电脉搏信号的VDT精神疲劳状态识别方法研究[D].[兰州理工大学硕士学位论文].兰州：兰州理工大学2011.6：48-49
[48]明东,田锡惠,杨春梅等.心率变异(HRV)信号的谱分析方法研究[J].北京生物医学工程.2001,20(12):251-255
[49] Akselrod S, Gordon D, Ubel F: Power spectrum analysis of heart ratefluctuation: a quantitative probe of beat-to-beat cardiovascular control. Science.1981，213,220--222
[50] Carolien S.E. Bulte, Sander W.M. Keet, Christa Boer, et al. Level of agreementbetween heart rate variability and pulse rate variability in healthy individuals[J]. European Journal of anesthesiology.2011,28(1):34-38.
[51]武艳华,黄纯,邢耀广,顾苏,潘华.基于AR模型的间谐波检测算法的研究[J].继电器.2006,34(2):41-55
[52]周芳，韩立岩.多传感器信息融合技术综述[J].遥测遥控.2006，27（3）：1-7
[53]杨万海.多传感器数据融合及其应用[M].西安电子科技大学出版社.2004
[54]邓魏,多传感器信息融合及其在农业的应用.农机化研究[J].2006.5:69-71
[55]鄢余武.分布异质传感器下的神经网络集成目标识别法.弹箭与制导学报[J].2007.3：112-114
[56] Yongjun Ma, Linqiang Zhan.Research on the Evaluation of Feature SelectionBased on SVM.Lecture Notes in Electrical Engineering.2012,1(133):407-414
[57] S. Aruna, S. P., Rajagopalan, L. V Nandakishore. Application of Gist SVM inCancer Detection[J]. Computer Science.2012,14:1203-1213
[58] Peipei Yin, Fuchun Sun, Chao Wang, Huaping Liu. An adaptive feature fusionframework for multi-class classification based on SVM [J]. Soft Comput.2008,12:685-691