awessds Springer-Verlag 1998

Transcription

1 Heart Vessels (1998) 13:79-86 Heart awessds Springer-Verlag 1998 Noninvasive measurement of aortic pressure waveform by ultrasound Hisashi Watanabe 1, Mie Kawai 1, Takahiro Sibata 1, Masatada Hara 1, Hiroshi Furuhata 2, and Seibu Mochizuki a 1Department of Internal Medicine (IV) and 2 Medical Engineering Laboratory, Jikei University School of Medicine, Nishi-shinbashi, Minato-ku, Tokyo , Japan Summary. At present, the aortic pressure (Pa) waveform can only be measured invasively. In this paper, we describe a new noninvasive method of measuring Pa. The aortic diameter (Da) pulse waveform was measured noninvasively from the suprasternal fossa using an echo-träcking system that was applied to the anterior and posterior aortic wall echoes. To eliminate viscoelastic distortion, the measured Da was converted to an estimate of Pa, named PÔ, using the stiffness parameter [3, which revealed the viscoelastic relationship between the vessel diameter and its internal pressure. P[3 was then compared with the Pa pattern that was measured directly. Eight patients with ischemic heart disease who had undergone cardiac catheterization were examined by this method. Results showed that (1) the Da and Pa waveforms were similar; (2) the P~ waveform resembled the Pa waveform more closely than did the Da waveforms for a single cardiac cycle (r = 0.970); and (3) in particular, P[3 resembled Pa most closely during the upslope phase of the ejection period (r = 0.996). Out results suggest that the Pa waveform can be accurately estimated from noninvasive measurements by this method. Key words" Pressure waveform - Aortic blood pressure - Aortic diameter - Ultrasound echo-tracking system - Noninvasive Introduetion The aortic pressure (Pa) waveform is one of the most important indices used in hemodynamic analyses. At Address correspondence to: H. Watanabe Received November 5, 1996; revision received September 14, 1998; accepted September 19, present, however, Pa can only be measured invasively, using a catheter-tip manometer. Currently, the peripheral pressure waveform can be measured noninvasively by using the ultrasonic quantitative flow measurement (QFM) system [1], the volume compensation method [2], and the tonometry method [3, 4]. When the peripheral pressure waveform is measured by tonometry, the central pressure waveform can be derived by a generalized transfer function (GTF) or individual transfer functions (ITFs) [5]. Clinical data derived using the GTF are useful for continuous monitoring of the blood pressure. However, the GTF is easily affected by changes in blood vessel properties caused by neurological influences, such as air cooling and neurological drugs. On the other hand, the peripheral vessel diameter can be used to estimate Pa. The carotid arterial diameter pulse waveform, obtained noninvasively by QFM, has been used to estimate Pa [6], but the nonlinear properties of the blood vessel wall and the nonlinear characteristics of propagation of the pulse wave introduce error into this estimate. Measuring the waveform noninvasively at a more proximal position would avoid these problems. We have developed a new method for measuring the aortic diameter (Da) pulse waveform and have used it to determine the aortic diameter pulse waveform (P~) compensated with the stiffness parameter [3. This report describes the validity of measuring P[3 noninvasively, by comparing it with the Pa waveform measured invasively. Materials and methods Patients Eight patients (7 males and 1 female) were evaluated during cardiac catheterization. All patients has ischemic heart disease without valvular disorders, and their average age was years.

2 80 H. Watanabe et al.: Noninvasive measurement of aortic pressure Capturing the aortic view echocardiographically The aortic view was obtained using a two-dimensional echocardiography system (Model SSH40A, Toshiba, Tokyo, Japan) operating with a 3.5-MHz transducer. The ultrasonic beam in the M mode was fixed in the optimal position for detecting Da so the beam was perpendicular to the vessel axis. A radio frequency (RF) signal output terminal was prepared for this study, and the RF signal in the M mode was red into the echotracking system described below. Da pulse waveforrn A block diagram and an outline of the principle of the ultrasonic echo-tracking system (USETS; Ultrasonic Arterial Displace Detector ME-502, ME Commercial, Tokyo, Japan) used in this study have already been reported [7, 8]. The USETS used the phase-locked loop method developed by Nakayama et al. [9, 10] and Hokanson et al. [11], which enabled vessel wall motion over a range of 20 ~m to 10mm to be measured continuously and automatically. The frequency response of the system was restricted according to the depth of the target vessel below the surface of the skin, and the detection depth and frequency response of the USETS were adjusted for measuring Da [8]. The RF signal was fed into the dual separated channels of the USETS. Each channel included an ultrasonic echo-tracking function enabling the USETS to automatically and continuously measure the distance between the echoes of the anterior and posterior walls of the aorta. The changes in this distance corresponded to the Da pulse waveform. Conversion of Da to an estimate of Pa using the stiffness parameter t3 The Da values were converted to P~ using Eq. 1. P[3 = Pdexp{[3*(D - Dd)/Dd} (1) where [3* = Dd/(Ds - Dd) ln(ps/pd); Ps = systolic pressure; Pd = diastolic pressure; Ds = vessel diameter at Ps; and Dd = vessel diameter at Pd. P = Poexp{[3(D - Do)/Do} (2) where P = blood pressure; D = vessel diameter; Po = standard blood pressure (100); Do = vessel diameter at Po; and [3 = stiffness parameter. Equation 1 represents the physical features of the vessel wall [12, 13]. It was obtained by modifying Eq. 2 which was devised by Hayashi et al. [14, 15]. Investigation of the most suitable measurernent position To locate the most appropriate USETS access point on the body, it is necessary to obtain the complete short- axis view of the aorta. To measure Da using this view, the ultrasonic beam must be perpendicular to the longitudinal axis of the vessel. It is well known that the suprasternal fossa, the aortic root, and Valsalva's sinus are all suitable access points on the body for obtaining an aortic short-axis view, so we compared the Da waveforms obtained from those three points with the Pa waveform measured directly and simultaneously, using a catheter-tip manometer, to select the best position. Measurement of Pa and Da, and calculation of Pfl Da, Pa, and an electrocardiogram were recorded simultaneously by a thermal alley recorder (Model WS-180G, Nippon Kohden, Tokyo, Japan). Pa was measured with a catheter-tip manometer (Model PC-370, Millar, Instrument, Houston, TX, USA). The recording paper speed was set at 100mm/s. The Da and Pa patterns were fed into a personal computer (PC9801 RA, NEC, Tokyo, Japan) with a digitizer (MITABLET-KD3838, Graphtec, Tokyo, Japan). The sampling time was set at 4-ms intervals, and the numerical data Ps, Pd, Ds, and Dd used in Eq. 1 were input using the keyboard. Using Eq. 1, P[3 was calculated from Da. To compare Da with Pa, we introduced the pressure waveform (Pw). Pw was obtained from the Da waveform pattern by substituting the values of Ps and Pd for Ds and Dd, respectively. The patterns of Pa, Pw, and P[3, and their Lissajious's figures, were displayed simultaneously. Furthermore, all the Lissajious's figures during one cardiac cycle and during the upslope period of the Pa waveform were superimposed, and the correlation coefficients of P~ - Pa and Pw - Pa were calculated. Results Optimal position for rneasurernent Figure I shows typical Da patterns of the aortic root and Valsalva's sinus obtained from the third intercostal area. Both patterns had blunt and distorted shapes, which differed from the Pa pattern (Fig. 2). The Da pattern of Valsalva's sinus was markedly different from the Pa pattern. The Da patterns of the aortic root and Valsalva's sinus in all patients showed the same distorted patterns as those shown in Fig. 1. Figure 3 shows the B-mode image of the aortic arch and its M-mode image obtained ffom the suprasternal fossa. The pulsations of the anterior and posterior walls throughout the cardiac cycle were captured. Figure 3 shows the simultaneously recorded Pa and Da patterns, measured by the USETS from the suprasternal fossa. The Da and Pa patterns of the aortic arch were similar, unlike those of the aortic root and Valsalva's sinus.

3 H. Watanabe et al.: Noninvasive measurement of aortic pressure 81 A Aortic-root B " Valsalva's sinus Valsatva's sinus I diameter Fig. 1. The ultrasonic echocardiogram and the vessel diameter pulse waveform patterns obtained by the echo-tracking system at the aortic root (A) and Valsalva's sinus (B) E C G ~ ~ - - ~ Aortlc p... form ] 1150(1011 igg Aortic diameter Pulse waveform 05 sec Fig. 2. Comparison between the aortic pressure pattern (upper trace) measured by catheter-tip manometer, and aortic diameter pulse waveform (lower trace) obtained by the echotracking system B mode M mode of Aortic arch anterior posterior wall * Fig. 3. Display of B-mode (upper) and M-mode (lower) echocardio- graphy of the aortic arch from the suprasternal fossa by ultrasonic echocardiography

4 82 H. Watanabe et al.: Noninvasive measurement of aortic pressure Pa... Pw i Pa =-- pfl P o 0_ 0 PO 150 lo0 j 5O ~150 ~-100 -O O O 123 5O I I Time(sec), 0,4 0, Fig. 4. Superimposed patterns of the aortic pressure waveform (Pa) and the aortic diameter pulse waveform (Pw) for case 1 Time(sec) i i i i i io Fig. 5. Superimposed patterns of Pa and the aortic diameter pulse waveform (P[3) compensated with the stiffness parameter [3 for case 1 Pw P,8 mrnhg rnrnhg.. / #1 J 001 ~,;~ i 00l ù,/" 50r y/ " 50./,, 50 1 O l O0 l 50 Pa Pa rnmhg Fig. 6. Lissajious's figures obtained from Pw/Pa and P[3/Pa for case 1. Pa, aortic pressure waveform; Pw, diameter pulse waveform; Pfl, diameter pulse waveform compensated with [3 Therefore, the USETS was applied to the suprasternal fossa of each subject for Da measurement. Comparison of Pa, Pw, and Pfi Figures 2, 4, 5, and 6 show the patterns obtained from the same male subject. Figure 4 shows the superimposed Pa and Pw patterns, which were similar, and Fig. 5 shows the superimposed P[3 and Pa patterns. Simple visual comparison of the two figures shows clearly that the P[3 pattern resembled the Pa pattern more closely than did the Pw pattern. This similarity was also evident from Lissajious's figures for P[3/Pa and Pw/Pa, shown in Fig. 6. The hemodynamic parameters for all subjects are listed in Table 1, and the individual correlation coefficients are presented in Table 2. The mean correlation coefficients of all subjects are shown in Fig. 7. The correlation coefficient for the relationship between P~ and Pa during one cardiac cycle was higher than that for Pw and Pa (r = vs r = 0.960, respectively). When only the data for the upslope phase of the ejection period were analyzed, the correlations improved. P[3 still resembled Pa more closely than did Pw (r = vs r = 0.993, respectively). Discussion Relationship between blood pressure and vascular diameter In the 1960s, it was discovered that vascular diameter pulse waveforms agreed well with the intravascular pressure waveforms in animal invasive experiments [16]. Furthermore, a strong association between the Da

5 H. Watanabe et al.: Noninvasive measurement of aortic pressure 83 Table 1. Clinical characteristics of the subjects Case Age (years) Sex Disease Blood pressure Maximum Minimum number () diameter (mm) diameter (mm) 1 51 M A.P. 155/ M A.P. 126/ M M.I. 169/ M M.I. 106/ M A.P. 152/ M M.I. 121/ F M.I. 106/ M M.I. 104/ Mean _+ SD 51.8 _ _+ 19.6/72.4 ± _ _ A.P., angina pectoris; M.I., myocardial infarction; [3, stiffness parameter Table 2. Correlation between Pw and Pa, P[3 and Pa Case One cardiac cycle Upsloping period of Pa number Comparison between Comparison between Comparison between Comparison between PW and Pa P6 and Pa PW and Pa P[3 and Pa r Regression r Regression r Regression equation equation equation y = 0.94x y = 0.97x y = 0.99x y = 0.89x y = 0.92x y = 0.92x y = 0.80x y = 0.83x y = 1.00x y = 0.97x y = 0.98x y = 0.93x y = 1.06x y = 1.06x y = 0.96x y = 0.81x y = 0.86x y x y = 0.95x y = 0.97x y = 1.10x y = 0.94x y = 0.96x y = 0.91x Pw, aortic diameter pulse waveform; Pa, aortic pressure waveform; P]3, pulse waveform compensated with [3 r Regression equation y = 1.01x - 1 y = 0.94x + 4 y = 1.01x - 3 y = 0.93x + 4 y = 0.98x + 4 y = 0.98x + 4 y= 1.10x- 10 y = 0.91x + 7 pulse and intraaortic pressure waveforms in patients who had undergone thoracotomy was noted [17]. Clinically, the noninvasive measurement of Da has not yet replaced invasive measurement of Pa because the available USETS did not have a frequency response suitable for measuring the rapid and large change in Da in the early systolic phase. Remarkably, a noninvasive estimation of the carotid arterial pressure waveform has been routinely performed by measuring the carotid arterial diameter [1]. This peripheral data, however, does not fully reflect the proximal Pa waveform. To obtain an accurate estimate of Pa by a noninvasive method, we developed a new USETS with a high-frequency response to aortic wall motion. In this study, we measured the Da of the aortic arch by a new noninvasive method using the improved USETS, and the Da pattern was similar to that of Pa. Optimal position for Da measurernent The suprasternal fossa was shown to be a suitable location for measuring, whereas measurements of Da made at the aortic root and Valsalva's sinus were distorted and did not closely match the pattern of Pa. The discrepancy stems from two factors: (1) aortic movement has three components, pulsation in the radial direction, displacement of the vessel axis, and fluctuation along the vessel axis, which is clearly seen in the longitudinal echocardiographic view of the aorta, and (2) the USETS does not have a three-dimensional tracking function for detecting vessel pulsation with the ultrasound beam. Accordingly, when measuring the Da at the aortic root and Valsalva's sinus, the tracking position of the vessels relative to the beam cannot be maintained continuously. Therefore, the shift of the tracking position during the cardiac cycle leads to distortion of

6 84 Pw 2OO H. Watanabe et al.: Noninvasive measurement of aortic pressure Pw r =0.960 Y=0.985X+9.! t 100 Pa one cardiac cycle O,.... r =0.993 Y=I.01X+0.6 I00 2{}0 rnmhg Pa upsloping phase P# P# j" I00 a~ Z," r =0.970 Y=0.989X A r =0.996 Y=I.00X-I ~" mrnhg Pa Pa one cardiac cycle upsloping phase Fig. 7. Lissajious's figures obtained from Pw/Pa and P[3/Pa of all cases during one cardiac cycle and during the upslope phase of the aortic pressm'e waveform

7 H. Watanabe et al.: Noninvasive measurement of aortic pressure 85 the Da pattern; that is, the pattern shows not only the pulsation component, but also an artifact caused by the fluctuation component. By contrast, the Da of the aortic arch coincided with the Pa measured using the catheter-tip manometer. This coincidence is attributable to the fact that aortic arch movement has only a pulsation component because the tissues surrounding the vessel reduced the other components of aortic movement and diminished the fluctuations. Out USETS could, therefore, measure the vessel pulsation component at the suprasternal fossa with adequate accuracy. Consequently, the comparison of the Da and Pa patterns led us to conclude that there are no positions other than the suprasternal fossa that are suitable for measuring the Da with a onedimensional tracking method like our USETS. ~150 q3-100 "O O o CO Pw PB Time(sec) i i i i I Fig. 8. Superimposed patterns of Pw and P[3 compensated with the stiffness parameter ~ for case 1 Correction by the stiffness parameter t3 The Da waveform measured at the aortic arch was similar to the Pa waveform, but they did not always coincide throughout the cardiac cycle. In particular, Da during the diastolic period showed a linear decremental waveform, whereas Pa normally showed an exponential downward concave curve. This discrepancy was thought to be due to the nonlinear elastic features of the vascular walls. Previous studies have shown that the relationship between the blood pressure and vessel diameter can be fitted to a uonlinear hysteresis curve. Equation 2, developed by Hayashi et al. [14, 15], expresses the static pressure-diameter relationship as an exponential function that included the stiffness parameter, [3. This formula is useful in clinical situations when a wide range of blood pressure is encountered, as was verified by their experimental study. Equation 2, however, requires that the vessel diameter is measured at a standard blood pressure of 100, making it impossible to apply the formula to noninvasive measurements. To overcome this problem, Asakawa et al. [12, 13] proposed a simplified formula (Eq. 1), in which the vessel diameter is obtained noninvasively. Equation 1 is based on the precondition that the vessel diameter reaches its maximum (Ds) and minimum (Dd) values when the blood pressure is at the systolic and diastolic pressure, respectively. This precondition was verified by the results of experiments on dogs, in which the parameters were measured invasively [16]. Equation 1 reduced the difference between the Pa and P~ patterns, as shown in Fig. 8. Prospects for the clinical use of noninvasive measurement of Pa The clinical application of our measurement procedure raises two problems concernius the calibration and the approximation of P~. The calibration of P[3 The maximum vessel pressure usually increases by approximately 10% from the aorta to the brachial artery due to the peaking phenomenon [18]. Therefore, the brachial cuff pressure may not be a suitable substitute for the aortic arch pressure in Eq. 2. To compensate for this deviation, mathematical relationships between the maximum and minimum brachial pressures and the Pa values, based on invasive clinical data, have been proposed [19, 20]. As an alternative calibration method, we used the tonometry method to measure the radial artery pressure. Chen reported that the central Pa can be accurately estimated by radial tonometry, using a GTF and ITFs [5]. Such proposals have the potential for error when used to obtain an instantaneous value for the aortic arch pressure, because this value varies during each heart beat, especially if the patient changes position. Therefore, our estimate of P[3 was not corrected for the pressure difference between the radial artery and the aorta, which poses a problem for the clinicm application of this method. Approximation of waveforms The difference between the P[3 and Pa waveforms was smaller than that between the Pw and Pa waveforms. In particular, the error in P[3 during systole was negligible, although the error during diastole was larger. It appears that P~ can be used for the noninvasive measurement of Pa, because it provided the closest approximation to Pa. On the other hand, there are several noninvasive methods that estimate the central blood pressure from peripheral blood pressure measurements, such as radial tonometry with a GTF or ITFs, and subclavian arterial pulse tracing. However, the accuracy of these estimates was confirmed using stationary subjects. We think that the

8 86 H. Watanabe et al.: Noninvasive measurement of aortic pressure clinical usefulness of our method should be assessed for various cardiovascular diseases, in comparison with other methods that use the peripheral arterial pressure to estimate Pa. Conclusions The pattern of the aortic diameter pulse waveform (Da), measured by a noninvasive USETS method, coincided well with that of the aortic pressure waveform (Pa), obtained using a catheter-tip manometer (r = 0.960). The pattern of the waveform that approximated to the aortic pressure waveform by incorporating a blood vessel stiffness parameter (P~) resembled Pa more closely than Da (r = 0.970). In particular, P[3 resembled Pa most closely during the upslope phase of the ejection period (P~ = 1.01 Pa , r = 0.996). Acknowledgrnents. This paper is dedicated to the late Kazuhiko Seki M.D., a very important member of our study group, who died just before this paper was submitted. We wish to express our most sincere appreciation of his diligent assistance and knowledgeable suggestions throughout the course of this study. References 1. Sugawara M (1987) Blood flow in the heart and large vessels. Med Prog Technol 12: Yamagoshi K, Rolfe P, Murphy C (1988) Current developments in non-invasive measurement of arterial blood pressure. J Biomed Eng 10: Karamanoglu M, O'Rourke MF, Avolio AP, Kelly RP (1993) An analysis of the relationship between central aortic and peripheral upper limb pressure in man. Eur Heart J 14: Chen CH, Ting CT, Nussbacher A, Nevo E, Kass DA, Pak P, Wang SP, Chang MS, Yin F (1996) Validation of carotid artery tonometry as a means of estimating augmentation index of ascending aortic pressure. Hypertension 27: Chen CH, Nevo E, Fetics B, Pak P, Ting CT, Yin F, Maughan L, Kass DA (1997) Estimation of central aortic pressure waveform by mathematical transformation of radial tonometry puressure - validation of generalized transfer function. Circulation 95: Sugawara M, Furuhata H, Aomi S, Kikkawa S, Yoshimura S, Caro CG (1982) A noninvasive method of measuring blood pressure wave (in Japanese). Iyoudensi to Seitaikougaku (Jpn J Med Electro Biol Eng) 20: Yosimura S, Kodaira K, Fujisiro K, Furuhata H (1981) A newly developed non-invasive technique for quantitative measurement of blood flow, with special reference to the measurement of carotid arterial blood flow. Jikeika Med J 28: Hara M, Seki K, Watanabe H, Miyashita Y, Takahashi I, Takayama K, Okamura T, Furuhata H (1990) Noninvasive measurement of aortic input impedance (in Japanese). Myakkangaku (J Jpn Coll Angiol) 30: Sato S (1971) Ultrasonic phase-locked echo-tracking system (in Japanese). Master's thesis, Sophia University 10. Nakayama K, Sato S (1973) Ultrasonic measurement of arterial wall movement a utilizing phase tracking system. Proc. Xth International Conferance on Medical and Biological Engineering Dresden, p Hokanson DE, Mozersky D J, Sumner DS, Strandness DE (1972) A phase locked echo tracking system for recording arterial diameter changes in vivo. J Appl Physiol 32: Asakawa T, Kawasaki T, Yagi S, Sinoyama S (1985) Noninvasive measurement of the mechanical property of arterial tree (in Japanese). Myakkangaku (J Jpn Coll Angiol) 25: Kawashaki K, Sasayama S, Yagi S, Asakawa T, Hirai T (1987) Noninvasive assessment of the age related changes in stiffness of major branches of the human arteries. Cardiovasc Res 21: Hayashi K, Sato M, Handa H, Moritake K (1974) Biomechanical study of the constitutive laws of vascular walls. Exp Mech 14: Hayashi K, Sato M, Niimi H, Handa H, Moritake K, Okumura A (1975) Analysis of constitutive laws of vascular walls by finite deformation therapy (in Japanese). Iyoudensi to Seitaikougaku (Jpn J Med Electro Biol Eng) 13: Lysle HP, Roderick EJ, Join P (1960) Mechanical properties of arteries in vivo. Circ Res 8: Greenfield JC Jr, Shapiro A (1962) Relation between pressure and diameter in the ascending aorta of man. Circ Res 10: McDonald DA (1974) Blood flow in arteries, 2nd edn. Edward Arnold, London, pp Borow KM, Newburger JW (1982) Noninvasive estimation of central aortic pressure using the oscillometric method for analyzing systemic artery pulsatile blood flow: comparative study of indirect systolic, diastolic, and mean brachial artery pressure with simultaneous direct ascending aortic pressure measurements. Am Heart J 103: Colan SD, Fujii A, Borow KM, MacPherson D, Sanders SP (1983) Noninvasive determination of systolic, diastolic, and end-systolic blood pressure in neonates, infants and young children: comparison with central aortic pressure measurements. Am J Cardiol 52: