Evaluation of Autonomic Nervous Activity Based on Arterial Wall Impedance and Heart Rate Variability

Transcription

1 Evauation of Autonomic Nervous Activity Based on Arteria Wa Impedance and Heart Rate Variabiity Abdugheni UTLU 1, enji SHIBA 1, Toshio TSUJI 1 Ryuji NAAMURA 2, Noboru SAEI 2, Masashi AWAMOTO 2 1 Graduate Schoo of Engineering, Hiroshima University, Higashi-Hiroshima, Japan Graduate Schoo of Biomedica Sciences, Hiroshima University, Hiroshima, Japan akutuk@bsys.hiroshima-u.ac.jp Abstract This paper proposes a new method for evauating the autonomic nervous activities by using the mechanica impedance of the arteria wa and heart rate variabiity. The cardiovascuar system is indispensabe to ife maintenance functions, and homeostasis is maintained by the autonomic nervous system. It is very important to give a diagnosis based on the autonomic nervous activities of the whoe body. The proposed method is evauated in surgica operations, where the mechanica impedance of the arteria wa is estimated from the arteria bood pressure and a puseoximetric photopethysmogram, and heart rate variabiity is estimated by eectrocardiogram RR interva spectra anaysis. From the experimenta resuts, we concuded that the proposed method is abe to estimate autonomic nerve activities corresponding to events during operations. eywords Mechanica impedance, Autonomic nerves, Arteria wa, Heart rate variabiity, Photopethysmogram. I. INTRODUCTION In medica treatment sites, a doctor needs to accuratey judge the patient's autonomic nervous conditions and take appropriate measures. Genericay, doctors used to judge patient's conditions during operations from vita signas and measures such as eectroencephaogram (EEG), eectrocardiogram (ECG) and bood pressure (BP) [1]. However, doctors must have expert knowedge and sufficient experience to catch subte changes in wave forms, infer causes, and make appropriate judgment. Particuary, in the case of emergency situations or during surgery, if a monitoring support system for cardiovascuar conditions becomes avaiabe, the medica staff wi be abe to more easiy and quicky identify patient's conditions. The cardiovascuar system is indispensabe to ife, and homeostasis is maintained by the autonomic nervous system (ANS). It is known that one part of the ANS based on heart activities is refected in fuctuations of the heart rate [2]. Many studies to date have been proposed for heart rate variabiity (HRV) by frequency domain anaysis and time domain anaysis [3-6]. Recent studies have aso used power spectrum anaysis to evauate ANS activities from changes in the bood pressure and heart rate during the evauation exercise [7-8]. However, these studies evauated with ony ANS activities appearing in the activity of the heart, and did not sufficienty evauate the entire cardiovascuar system. On the other hand, some researchers have tried to describe the detaied characteristics of vascuar smooth musces, wherein vascuar tone (hardness) represents an index of ANS activity of the arteria wa [9-11]. Aso, appying the skeeta musce impedance estimation method [13], the authors modeed the dynamic characteristics of the human arteria wa by empoying mechanica impedance, incuding stiffness, viscosity and inertia to estimate the changing beat-to-beat conditions of bood vesses and ascertained the vascuar conditions in the physician s surgica actions as impedance changes [12-15]. However, the deveoped system ony evauated the ANS activity of the periphera vesse, and was not enough to understand the autonomic nervous activity of the entire cardiovascuar system. The present paper proposes a new monitoring system to comprehensivey evauate ANS activity on the cardiovascuar system by simutaneousy anayzing the arteria wa impedance and the HRV, and aso dispaying the bispectra index (BIS) estimated from EEG. In the proposed system, the ANS activity of the heart is estimated by HRV, and the ANS activity of bood vesse is estimated by arteria wa impedance. The proposed system aims to deveop an on-ine monitoring (support) system in surgica operations, which can estimate ANS activity from beat-to-beat. Using the proposed system, we monitor ANS activity during endoscopic transthoracic sympathectomy surgery and assess the vaidity of the proposed method. II. CONFIGURATION OF MONITORING SYSTEM A structure of the proposed diagnosis ANS activity support system is shown in Fig. 1. The proposed system consists of three parts, I. Signa measurement, II. Parameter estimation, III. Dispay. In the monitoring support system, parts II and III are constructed with a graphica dispay screen using LabVIEW (Nationa Instruments Corp.). The methods of preprocessing and estimating ANS activity used in a prototype are described beow /7/$ IEEE 469

2 Subject Bioogica signas EEG A-2 BSS-98 ECG IBP PLS Processing BIS ECG IBP PLS R-R HF LF/HF ~ Synchronizing Dispay Measured signas and estimated parameters BIS, ECG, IBP, PLS R-R, HF, LF/HF, ~ Tissue Arteria wa B M σ ε I II III Measuring signas Parameters estimation Monitoring of autonomic nervous activity Tunica intima Fig. 1 System configuration Fig. 2 The arteria wa impedance mode A. Signa Measurement ECG, BP and a puseoximetric photopethysmogram (PLS) are measured at 125 Hz simutaneousy and stored on a persona computer by using Transmission Contro Protoco (TCP). The BP is measured through a catheter (24 gauge) paced in the eft radia artery, and the PLS is measured with the ipsiatera thumb (BSS-98, NIHON OHDEN Corp.). The bispectra index (BIS) was recorded using the Aspect A- 2 EEG monitor (Aspect Medica Systems Inc.) and BIS vaues were transferred in 5 second intervas and stored on a persona computer. B. Estimation and Dispay of Autonomic Nervous Activity 1) Estimation of Arteria Wa Impedance: For extracting vascuar features incuded in the measured signas, the arteria wa impedance is estimated [14]. Fig.2 iustrates the proposed impedance mode of the arteria wa. This mode represents ony the characteristics of the arteria wa in the arbitrary radius direction. The impedance characteristic can be defined as a reation between an exerted force and a dispacement of the arteria wa, expressed in the foowing equation: dσ = Md & ε + Bd & ε + d ε (1) where σ (t) is the stress exerted on the arteria wa by bood pressure; M, B, and represent the inertia, viscosity, and stiffness; ε (t), ε& (t), and ε&(t & ) are the strain, strain veocity, and strain acceeration of the wa; the coefficient d means the variation from the time t, dσ = σ( t) σ( t ), dε = ε ε ( t ) ; and t denotes the time when the R wave appeared in each ECG recognition cyce [15]. To estimate the impedance parameters given in equation (1), it is necessary to measure σ (t) and ε (t). Assuming that the stress σ (t) is proportiona to the direct arteria pressure P b (t), the foowing equation can be obtained: ECG (t) 11.5 RR interva where k f is a proportiona constant. On the other hand, the vascuar strain ε (t) is quite difficut to measure directy. Then, a pethysmogram is utiized instead of ε (t) as foows: ε = k P (3) p where P (t) is the measured pethysmogram, and k p is a proportiona constant [8]. The stress exerted on the arteria wa is expressed by the direct arteria pressure P b (t) given by equation (2), and the strain of the arteria wa ε (t) is represented by the pethysmogram P (t) in (3). It has been shown that inertia M is very sma and can be ignored in the transversa arteries [16]. Then, the arteria wa impedance parameters can be estimated by using the east square method from the measured signas P b (t) and (t), which gives rise to foowing equation: P R P Q S Time [s] Fig. 3 ECG wave dp = B ~ dp& ~ dp (4) b + where dpb = Pb Pb ; dp = P P ; dp & ( t ) = P & ( t ) P & ( t ) ; dp& = P&& P&&. B ~ and ~ correspond to the viscoeastic properties of the arteria wa respectivey. T σ = k P (2) f b 47

3 2) Heart Rate Variabiity Estimation: Heart rate rhythm is adjusted by ANS activity. The variation of beat-to-beat intervas referred to heart rate variabiity (HRV), can usuay be cacuated by anayzing the time series of R-R intervas from ECG in Fig. 3. Various HRV spectra anayses have been proposed using the Fast Fourier Transform (FFT) and Waveet Transform (WT) anaysis [3-5]. The principe of spectra anaysis is based on the fact that the sympathetic nervous system (SNS) and parasympathetic nervous system (PNS) activities refect on the heart rate (R-R interva) variabiity in specific frequency bands. Generay, it is known from the bocking drugs experiment of the nervous system that the ow frequency components (LF:.4~.15Hz) refect both SNS and PNS activity, and that the high frequency components (HF:.15~.5Hz) refect the PNS activity. The ratio LF/HF refects the SNS activity [4-5]. In this study, the cardiac ANS activity is estimated by using WT (high time resoution) based power spectra anaysis, and HF and LF/HF components are computed. We utiized Fucet software (Dainippon-Sumitomo pharma Co. Japan) for the HRV anaysis [6]. 3) Dispay: The dispay of the deveoped system is given in Fig. 4. In the upper eft portion of the screen, the measured eectrocardiogram (ECG), invasive arteria pressure (IBP) and photopethysmogram (PLS) are shown as the measured bioogica signas. The upper right portion of the screen shows the Lissajous figure, where the of a horizonta axis is the IBP wave and the vertica axis is the PLS wave, so that the incination of the Lissajous figure, which is the ratio of the ampitudes of the pethysmogram and arteria pressure, gives an approximation of bood vesse compiance [9]. At the bottom of the screen, the estimated 5 parameters (BIS, R-R, HF, LF/HF, ~ ) represent an index of awakening of the brain and the ANS activity of the heart and arteria wa. In this system, the changes in a patient's ANS activity on the cardiovascuar system can be understood both visuay and easiy. The medica doctors can predict whether a patient's condition wi take a turn for the better or become more and more dangerous from the changes in the signas. Fig. 4 The parameters on the dispay screen image START START Measuring one samping data of ECG,IBP, and PLS Measuring one samping data of EEG NO ECG, IBP, PLS Reading the each data BIS vaue Dispay of the measured 32 samping data of ECG, IBP, and PLS NO End the measuring of 32 samping data? Process A Eimination of noise by fiters Saving the 32 samping data of ECG, IBP, and PLS Saving the each 5 sec cacuated data of EEG NO = 5 sec? Detection of R wave? ECG, IBP, PLS BIS vaue Estimation of impedance parameters and dispay the estimation resuts Process B (a) Measuring the bioogica signas (b) Signa processing Fig. 5 Fowchart of the signa processing 471

4 4) Process Fowchart: Fig. 5(a) expains the measuring of the bioogica signa for communication of the persona computer and the bedside monitor. The bedside monitor measures the eectrocardiogram, arteria pressure, and photopethysmogram simutaneousy (Fig. 5(a)). The samping time is 8 ms. The measured bioogica signas are stored in the bedside monitor and the data are transferred a at once using TCP when 32 samping data are stored. The transfer time is about.2 ms, which is sufficienty short compared with the samping time. Fig. 6 demonstrates an exampe of the measured bioogica signas. This figure pots the eectrocardiogram, arteria pressure, and photopethysmogram. The BIS vaue is cacuated by the EEG monitor, which are transferred in every 5 seconds and stored on a persona computer. Fig. 5(b) expains the estimation of parameters on the persona computer. The measured bioogica signas are read in every 32 samping data, and then dispayed in the upper eft portion of the screen (process A). Because the data are affected by some artifacts, such as ight and mechanica stimuation on the patient's hand, the arteria pressure and photopethysmogram are preprocessed using digita fiters. The invasive arteria pressure were fitered out through a second-order infinite impuse response (IIR) owpass fiter with the cutoff frequency of 6 Hz and a first-order IIR high-pass fiter with a cutoff frequency of.3 Hz, respectivey, and the photopethysmogram is fitered through an eighth-order finite impuse response (FIR) ow-pass fiter with the cutoff frequency of 15 Hz and a first-order IIR high-pass fiter with the cutoff frequency of.3 Hz. After the fitering process, the R wave in each ECG recognition cyce is detected (process A end). If the R wave is detected, then processing is begun to estimate (process B) each parameter (R-R, HF, LF/HF, ~ ). When the R wave is detected, the arteria pressure and pethysmogram, which are measured data from the previous R wave to the detected R wave, are substituted in equation (4) and the impedance parameters are estimated. Because the previous R-R interva is estabished each time an R wave is detected, the beat-to-beat arteria wa impedance parameters ~ can be estimated. The estimation resuts are dispayed at the bottom of the screen (resuts ~ ) in Fig. 4. At the same time, waveet transform (high time resoution) based HRV anaysis is used to estimate the cardiac ANS activity (R-R, HF, LF/HF). The estimation resuts are dispayed at the bottom of the screen (resuts R-R, HF, LF/HF) in Fig. 4 (process B end). As for the estimation time of the parameters (BIS, R-R interva, HF, LF/HF, ~ ) showed in Fig.4, R-R and ~ are renewed from beat-to-beat, and parameters BIS, HF, LF/HF are synchronized every second and then renewed on the screen. The medica doctors can identify whether the patient's condition is stabe from the bioogica signas, arteria wa impedance and cardiac ANS activity. Aso, they can be judged whether anesthesia is adequatey taking effect or not. III. MONITORING EXPERIMENT To evauate the vaidity of the proposed method, the ANS activity during the surgica operations was evauated by using ECG [V] IBP [mmhg] PLS [%] Time [s] Fig. 6 Exampes of the bioogica signas off-ine anaysis. The proposed system aims towards an on-ine monitoring that can be appied to surgica operations. As a hafway stage of system deveopment, the bioogica signas measurement was taken on-ine, and the signas anaysis was performed taken off-ine. The operations on the endoscopic transthoracic sympathectomy for hyperhidrosis with four patients were used for off-ine anaysis. The endoscopic transthoracic sympathectomy is a surgery for hyperhidrosis patients, whose sympathetic nerves are in a hyperfunction state, and the patients perspire copiousy in the pams and armpits compared to ordinary peope. In this operation, the sympathetic nerve running aong the spine is cipped to stop perspiration [17]. When the sympathetic nerve is cipped, it is expected the bood vesses become compiant on the spot. In this operation, ECG, BP, PLS, EEG are measured simutaneousy by using the ipsiatera thumb and EEG monitor. The BP was measured through a 24 gauge catheter paced in the eft radia artery, and the PLS was measured from both the eft and right thumbs. A persona computer LATITUDE/61 (DELL, Pentium IV 2.GHz) was used for the monitoring system. IV. RESULTS AND DISCUSSION In order to assess the vaidity of the proposed method, the data of four patients (patient A, B, C, D) with correct interruption of the sympathetic nerves were anaysed. Fig. 7 shows an exampe of experimenta resuts of the estimated ANS activities during surgery. In order from the top, the BIS, R-R interva variabiity, HF, LF/HF, and the periphera ANS activities ~ are described. The right and eft side sympathetic nerves of the patient were cipped at the dotted ines A and B. In the figure, the timing of the induction and the cessation (wearing off) of anesthesia are indicated as (I) and (IV). The surgeon stimuated the patient's tissues by identifying the sympathetic nerve from right to the eft at the area (II) and (III)

5 It can be seen that, as for the timing on the induction of anesthesia (I) and cessation of anesthesia (IV) area, the sympathetic nerves activities (LF/HF) decreased and increased, respectivey. When these painfu stimui were appied (E1, E2), periphera vascuar tone ~ increased because sensory information was transmitted from the medua spinais to the brain, and periphera vasoconstriction and musce contraction occurred. After the sympathetic nerve was cipped, the bood vesses became markedy compiant (F1, F2) and ~ decreased because sympathetic nerve activities coud not be transmitted to the periphera part. Furthermore, it can be understood that because the surgeon stimuated and cipped the patient's sympathetic nerve. It shoud be noted that the LF/HF activities increased before the cipping (C1, C2) and after the cipping (D1, D2). Fig. 8 shows the comparative resuts of the mean vaues and the standard deviations for 4 patients in 18 seconds of data on the resting (norma) state (i), before the cipping (ii) on the right side, after the cipping (iii) on the right side, before the cipping (iv) on the eft side, and before the cipping (v) on the eft side. Because the sympathetic nerve was cipped adequatey, after the cipping ((iii) and (v)) ~ recovered to the resting eves (i) after the nerve was cipped. Aso, it can be understood from the HRV resuts that LF/HF activities are increased from (ii), (iv) to (iii), (v). This coud be caused by the cipping, the effects of which appeared in the cardiac ANS activity. Consequenty, it can be seen that, from LF/HF and ~, we can monitor the patient s cardiac and periphera autonomic nervous activities. It can aso be seen that anesthesia depth was obtained based on comparison of LF/HF and ~ with the BIS index. V. CONCLUSIONS We have deveoped a monitoring system for estimating ANS activity by simutaneousy anayzing the arteria wa impedance and the heart rate variabiity, and have ascertained the vaidity of the proposed method during the surgery. Our resuts show that the estimated ANS activities correspond to the events during operations. Future research wi be directed to test the onine version of the proposed system and the number of subjects wi be increased. It wi aso be ascertained for a wide variety of uses in the genera environment. ACNOWLEDGMENT The authors woud ike to thank Mr. T.Ukawa of the NIHON OHDEN Co., Ltd for making avaiabe cinica data for this research. This work was supported by the knowedge Custer for Scientific Research of Japan Society for the Promotion of Science. BIS R-R interva [s] HF LF/HF ~ [mmhg/%] HF LF/HF Left Right ~ [mmhg/%] Left Right (Ⅰ) (Ⅱ) (Ⅳ) Time [s] 5 Fig. 7 Exampes of the estimated parameters (Patient A) (ⅰ) Resting Fig.8 Estimated resuts of autonomic nervous activities REFERENCES [1] T.umazawa, O.Yugi, H.Furuya, Standart anesthesia, pp , 22, (in Japanease). [2] H.Inoue, Cardiovascuar disease and autonomic nervous function, pp. 5-1, 21, (in Japanease). [3] Akserod.S, Gordon.D, Ube.FA, Shannon.DC, Berger.AC, Cohen.RJ, Power spectrum anaysis of heart rate fuctuation: a quantitative probe of beat-to beat cardiovascuar contro, Science, Vo. 1, pp , [4] Yamamoto.Y, Hughson.RL, Peterson JC, Autonomic contro of heart rate during exercise studied by heart rate variabiity spectra anaysis, J App Physio, Vo. 71, pp , [5] Task Force of the European Society of Cardioogy and the North American Society of Pacing and Eectrophysioogy, Heart rate variabiity: standards of measurement, physioogica interpretation, and cinica use, Circuation, Vo. 93, pp , A (Ⅲ) B C1 D1 C2 D2 E1 F1 E2 F2 (ⅱ) (ⅲ) (ⅳ) (ⅴ) Before After Before After Vesse Heart Brain :Patient A :Patient B :Patient C :Patient D Time[s] 473

6 [6] Bi-Hua Tan, et a, Waveet transform anaysis of heart rate variabiity to assess the autonomic changes associated with spontaneous coronary spasm of variant angina, J Eectrovcardioogy, Esevier, Vo.36(2), pp , 23. [7] Tayor J.A and D. L. ECBERG, Fundamenta reations between short-term R-R interva and arteria pressure osciations in humans, Circuation, Vo. 93, pp , [8] R. Zhang,. Iwasaki, J. H. Zuckerman, et a, Mechanism of bood pressure and R-R variabiity insights from gangion bockade in humans, Journa of Physioogy, Vo. 543 (1), pp , 22. [9] R.L.Armentano, A.Simon, J.Levenson, N.P.Chau, J.L.Megnien, and R. Piche, Mechanica pressure versus intrinsic effects of hypertension on arge arteries in humans, Hypertension, Vo. 18(5), pp , [1] J.G.Barra, R.L.Armentano, J.Levenson et,a, Assessment of Smooth Musce Contribution to Descending Thoracic Aortic Eastic Mechanics in Conscious Dogs, Circ. Res., Vo. 73, pp , [11] J.G.Barra, S.Graf, and R.L.Armentano, Beneficia Effect of Aortic Smooth Musce Energy Dissipation in Conscious Dogs, in Proc. Word Congress on Medica Physics and Biomedica Engineering, pp. 586, 23. [12] N.Saeki, M.awamoto, and O.Yuge, Quantitative view of periphera circuation, Critica Care Medicine, 28-12, A62(supp), 2. [13] T.Tsuji, P.G.Morasso et a, Human hand impedance characteristics during maintained posture, Bioogica Cybernetics, Vo. 72(6), pp , [14] A.Sakane, T.Tsuji, Y.Tanaka, N.Saeki, M.awamoto, Monitoring of Vascuar Conditions Using Pethysmogram, The Society of Instrument and Contro Engineers, Vo. 4(12), pp , 24 (in Japanease). [15] A.Sakane, T.Tsuji, N.Saeki et a, Discrimination of Vascuar Conditions Using a Probabiistic Neura Network, Journa of Robotics and Mechatronics, Vo. 16(2), pp , 24. [16] W.W.Nichos, et a, McDonad's Bood Fow in Arteries: Theoretica Experimenta and Cinica Principes, Arnod, 4th ed, London, [17] C.Drott, G.Gothberg, and G.Caes, Endoscopic transthoracic sympathectomy: an efficient and safe method for the treatment of hyperhidrosis, Journa of the American Academy of Dermatoogy, Vo. 33(1), pp ,