Time-Frequency Analysis of Heart Rate Variability and Investigation of the Function of Autonomic Nervous System

• Main Authors: Chan, Hsiao-Lung, 詹曉龍
• Other Authors: Wu Chien-Ping, Lin Jiunn-Lee
• Format: Others
• Language: zh-TW
• Published: 1997
• Online Access: http://ndltd.ncl.edu.tw/handle/90238816182935789196

博士 === 國立臺灣大學 === 電機工程學系 === 85 === Heart rate variability (HRV) is mainly due to the modulation of theautonomic inputs to the heart, and the analysis of HRV also provides aprobe to understand the autonomic function. Recently, severalphysiological and clinical studies have demonstrated that the spectralcomponents of HRV at low frequency and high frequency can reveal thefunction of two branches of the autonomic nervous system - sympatheticand parasympathetic nervous systems, respectively. The dysfunction ofthe cardiovascular system will also deteriorate the normal function ofthe autonomic nervous system and reduce HRV.Most studies dealing with the spectral analysis of HRV is based on theFourier spectral analysis. However, the characteristic of HRV ischanged quickly in some situations, e.g. postural change,pharmacological influence, anesthetic events, etc. So, the Fourierspectral analysis is unable to catch the transient characteristic dueto its underlying assumption of stationarity of data. Although theadvanced spectral analysis method, time-frequency distribution canreveal the time-related spectral components of the nonstationarysignals, the special characteristics of HRV (like 1/f spectralcharacteristic, very low frequency trend, transient change, etc.) mayaffect the time-frequency representations of HRV. In thisdissertation, a new time-frequency distribution is proposed. Theconcept of the proposed method is to change the spectralcharacteristic of HRV by f^alpha pre-filtering before time-frequencyanalysis, and the change of spectral energy is recovered bymultiplying a frequency compensation function 1/f^alpha at finalstage. Based on the tests by simulation and real HRV signals, theproposed method is superior than other time-frequency distributions inrevealing the time-related frequency components of HRV and estimatingthe spectral energy qualitatively and quantitatively.Based on the proposed time-frequency distribution, the spectralcharacteristics of HRV under general anesthesia (awake -> thiopentoneinduction -> tracheal intubation -> isoflurane anesthesia maintenance)is investigated. Our result shows a reduction of low-frequency andhigh- frequency heart rate fluctuations from awake to thiopentoneanesthesia, and low- frequency fluctuation is further reduced inisoflurane maintenance. In addition, instead of the elevation oflow- frequency spectral power, the sympathetic activation due totracheal intubation reflects on the increase of the ratio oflow-frequency to high-frequency spectral power.The autonomic imbalance due to the sympathetic over-activation is oneof the syndromes of severe congestive heart failure. The HRV ofcongestive heart failure usually has two special characteristics: (1)a very low heart rate fluctuation, (2) a very low frequency heart rateoscillation, named as Cheyne- Stokes respiration (CSR). The former ismainly owing to the suppression of the autonomic nervous system, andthe latter is probably due to that the interaction between theautonomic nervous system and cardiovascular system is controlled by aspecific mechanism. Clinically, the low-dose beta-blocker therapy canimprove the autonomic imbalance and the cardiovascular function.Nevertheless, it is not clear that the change of HRV, its relatedautonomic function and cardiac performance after low-dose beta-blockertherapy. In this dissertation, the change of the characteristic of HRVis investigated after low-dose beta-blocker therapy by 24-hour HRVstudy. Our result demonstrates that the low- frequency andhigh-frequency heart rate fluctuations are increased as the elevationof cardiac function, which implies the recovery of the autonomicfunction. The augmentation of low-frequency spectral power is largerthan that of high frequency, and it might be due to the recovery ofbaroreceptor sensitivity. In addition, we employ time- frequencydistribution to interpret the time-related spectral characteristics ofCSR, and develop a new method to discriminate CSR based on the theoryof pattern recognition. Our result reveals that the occurrence of CSRhas been reduced as the recovery of cardiac and autonomic functionsafter low-dose beta-blocker therapy.