In vivo comparison between two tip pressure transducer systems

InternationalJournal of Clinical Monitoringand Computing 12: 77-83, 1995. (g) 1995KluwerAcademicPublishers. Printedin the Netherlands.

77

In vivo comparison between two tip pressure transducer systems A n d r 6 E. A u b e r t , M a t t y V r o l i x 1, H i l a i r e D e G e e s t & F r a n s V a n d e W e f t Department of Cardiology, University Hospital Gasthuisberg, Catholic University Leuven, Belgium; Lpresent address: St. Jansziekenhuis, Gent, Belgium Accepted5 April 1995

Key words: tip catheter, pressure transducer, left ventricular pressure, cross correlation

Abstract Experimental findings are presented of an in vivo comparison between a Sentron catheter and another tip transducer manometer: a Millar microtip catheter. Both catheters have been used simultaneously in the left ventricle of dogs. Pressure variations were elicited by drug infusion. Pressure values and derivatives obtained from both systems were compared. A cross correlation between episodes of the two pressures was computed. Results from this study showed good correlation between left ventricular systolic pressure measured with both manometers (R = 0.992, p < 0.0001), end-diastolic pressure (R = 0.809, p < 0.0001) and between first derivatives: positive derivative (R = 0.993, p < 0.0001) and negative (R = 0.634, p < 0.0001). The mean cross correlation between both pressure signals was 0.61 + 0.04. In the frequency domain no statistical difference was found between the location of the maxima of the peaks. It is concluded that a Sentron manometer can be a valid alternative, at a reasonable price, to a cheaper, though less accurate fluid filled catheter and a more expensive 'golden standard' microtip catheter.

Introduction The general objective of physiologic pressure measurements is to record pressure variations faithfully, usually in permanent analog form. The process will involve a transducer capable of changing mechanical pressure signals into a form that is convenient, which normally requires conversion to an electrical signal. The ideal unit would be a small, safe, stable, reliable, nonhostile to a physiologic environment and inexpensive. Moreover it should be easy to introduce and capable of remaining in the desired location for a longer period of time. Catheter pressure transducers are not new. An air balloon type was first used by Marcy [1] more than 100 years ago and a resistive type by Grunbaum [2J at the turn of the century. In the sixties Telco tip catheters [3] were used. These transducers were based on an inductive principle. A widely used tip catheter was described by Millar and Baker [4]. In this study this type was used as a reference.

In the catherization laboratory intracardiac or vascular pressures are most often measured by means of the conventional catheter-manometer system consisting of a fluid filled catheter, connecting stop cocks and an external pressure transducer. This system is relatively cheap and rugged. However, its performance is limited due to the transmission characteristics of the catheter fluid filled column and in the deflection of the transducer strain gauge. Especially for the reliable determination of first derivatives of left ventricular pressure micromanometer devices rather than water filled catheter systems [5] should be used. Recently a new tip catheter manometer (Sentron) was introduced. It is cheaper and is claimed to have similar characteristics as other tip manometers. It was the purpose of this study to compare data obtained from this tip catheter manometer with a Millar type.

78 Experimental methods Animal preparations Experiments were performed in 4 mongrel dogs weighing 18 to 26 kg. Anesthesia was induced with fluanisone (Hypnorm) (0.5 ml/kg body weight s.c.) and maintained with sodium pentobarbital (Nembutal). The dogs were ventilated with a mixture of 50% oxygen and 50% room air by means of a Bird respirator through a cuffed endotracheal tube. Heart rate was between 70 and 90 beats/min. Lisinopril, which is an angiotensin converting enzyme (ACE) inhibitor, was injected as a continuous intracoronary infusion at various concentrations in order to elicit pressure variations. All animals received human care in compliance with the 'Principles of Laboratory Animal Care' [6]. Pressure measurements Left heart catheterization was performed from the right and left femoral approach. A left coronary angiogram was performed from the left carotid artery. All pressures were referenced to atmospheric pressure at the level of the midchest and were recorded during short periods of apnea. The pressure in the left ventricle was measured with a high-fidelity micromanometer (Millar Instruments, Houston, TX), used as a reference and with another micromanometer tip catheter (Sentron, Roden, The Netherlands). Prior to insertion both pressure sensors were emerged in a 37 ~ C 0.9% saline solution to minimize temperature drift. Mechanical calibration was performed for both pressure transducers in a pressure chamber (H. Gauer) by applying pressures from zero to 100 mmHg and adjusting amplification and matched against luminal pressure. The catheters were positioned under fluoroscopic control at a distance of approximately 2 cm. Both pressure signals were recorded simultaneously on an ink-jet writer (Siemens-Elema) and directly digitally on a personal computer. Pressure sensor The Sentron pressure sensor catheter system consists of a catheter and an electronic interface. The catheter (Type No 811-160) contains a Silicon microchip transducer near the tip. The chip comprises 4 piezo resistors (full Wheatstone bridge) integrated on a micromachined Silicon membrane (370/1 z surface, 8 # thick). The pressure sensitivity of the (uncompensated) chip

ranges from 9.5 to 20 #V/V/mmHg, a range of - 50 to 300 mmHg and a thermal stability of 0.1 mmHg/~ C. The catheter connector comprises a digital memory in the form of a PROM that contains all data required by the Sentron interface to compensation/calibration of pressure sensitivity and temperature effects. The catheter has an inner lumen for guide wire passage and contrast injections. The Sentron interface (Type No 811-000/A) is a microprocessor controlled signal conditioner that enables easy calibration of the sensor (zeroing) and generates reference signals for the calibration of peripheral equipment (pressure sensors). The Sentron system has a frequency bandwidth of 0 180 Hz. The pressure sensitivity of the Millar transducer (at 25 ~ C) is 5 #V/V/mmHg, a range of + 300 mmHg, a thermal stability of + 0.1 mmHg/~ C and it has a bandwidth of 0 to 20000 Hz. The Millar sensor is also a piezo resistive transducer. Data analysis All signals were digitally recorded, in episodes of 10 s, on a Vectra/RS personal computer (Hewlett Packard, Stanford, CA) with an Inte180386 processor (20 MHz), and Intel 80387A coprocessor and a 150 Mb hard disk. The analog/digital conversion was performed with a Data Translation DT2821 data acquisition board housed in one of the computer expansion slots. The gain was set for full-scale resolution of the A/D conversion board. Data acquisition at a sampling frequency of 1 kHz, graphic display of the signals and analysis programs were written in ASYST (Keithley, Rochester, NY). At optimal gain and full scale registration of a pressure signal of 150 mmHg, the amplitude resolution was 0.5 mmHg. In total 42 episodes were recorded, equally distributed over the 4 experiments. All data computed from the pressure tracings are mean values from 4 to 7 heart beats. Differentiation was performed by interpolating a polynomial through consecutive data points and then differentiating the polynomial. The degree of the polynomial is two by default. Power spectra were obtained with a digital fast Fourier transform method (DFP-T). Statistical analysis All data are reported as mean • SD. Statistical significance was set at values ofp < 0.05 and was obtained by

79

Systolic pressure

Table I. Coefficients and correlation coefficient for linear curve fits between systolic pressure (LVP) in mmHg, end-diastolic pressure (EDP) in mmHg, maximal positive derivative (dp/dt)+ in mmHg/s and minimal negative derivative (dp/dt)- in mmHg/s for the Sentron (S) and Millar (M) tip catheters. Y$

=

LVP EDP (dp/dt)+ (dp/dt)EDP S =

A] XM

R

p value

1.75

0.98

0.992

0.0001

6.22 - 93.84 - 63.14

0.42 1.01 1.11

0.809 0.993 0.634

0.0001 0.0001 0.0001

1.45 EDPM

0.916

0.05 (EDP m)2 0.0001

A0

+

2.09 +

200

hE

E

I00

R: correlation coefficient (N = 42). A parabolic curve fit is also given for EDE 0

a Student's test of paired data. Linear curve fit between data from both catheters were obtained. The cross correlation function (after removing of the DC component) between the two sets of pressure data was computed. The cross correlation function of two sets of data [7] describes the general dependence of the values of one set of data on the other. The cross correlation function of the time history records x(t) and y(t) can be described as:

100

0

200

Millar (mmHg)

Fig. 1. Linear correlation between maximal systolic pressures.

EDP 20

T

1/T j x(t)y(t

+ -,-)dr

16

o ++ 32 E E T

---+ ( 2 0

+

_+_ .= ~%-'" +

i2 +

c o

+

/

c

This function was normalized by dividing with the mean square amplitude of the signals x(t) and y(t);

T o

09

J'4

+#+"

} o

The resulting function will only take values between - 1 and 1. An absolute value closer to ] will indicate a better correlation. This plot will display sharp peaks that indicate the existence of correlation between x(t) and y(t) for specific time displacements. Hence a time difference between the two signals can be established directly by noting the time displacement associated with an observed peak in the cross correlogram.

0

4

8

12

16

20

Millar (mrnHg)

Fig. 2A. Linear correlation between end-diastolic pressures (EDP).

Results Pressure data were obtained in time and frequency domain. In the time domain discrete measurements

80

(dp/dt)-

EDP

4000

20

3200 16

+++44- + "IE E

++ ~ +

12

+

~----._~++ +-~+

E E

+

2400

o=

t"

c

?. E

1600

r

800 i

0

1000

.

i

L

2000

3000

4000

Millar (turn Hg/s) 0

8

4

Mil|ar

12

16

20

(mmllg)

Fig. 4.

Linear correlation between minimal negative derivatives

(dp/dt)-. Parabolic correlation between end-diastolic pressures (EDP) (same data points as in Fig. 2a).

Fig. 2B.

xE4

(dp/dt)+ 4.90

4000 21.50

3000

w 2.+.+

-I-

E E

2000

E

Y~ 1000 f (Hz)

,~E~

Fig. 5A. Powerspeclrum of left ventricular pressure obtained from

0 0

800

1600

2400

3200

4000

Millar catheter.

Millar (mmHo/s)

Pressures Fig. 3.

Linear correlation between maximal positive derivatives

(dp/dt)+.

were obtained and frequency components were calculated with digital fast Fourier transform.

During different episodes pressure variations were elicited and peak systolic and end-diastolic (EDP) pressures measured. Figure 1 shows a comparison of systolic pressures, pressure ranges for the Millar catheter was: 81 to 149 m m H g and for the Sentron catheter: 81 to 142 mmHg. Mean values for all 42 points are

81 x

E

4

| ~

.

.

.

.

.

. . . . . . . . .

. . . . . . . . .

.. . . . . . . .

9

. . . . . . .

,

. . . . . . . .

9 . . . . . . . .

i i ! i i i

4.9Ie

. . . . . . . . .

i ........

i........

i .......

i .......

~ .......

9.

.

.

.

.

:. . . . . . . .

Table 1. A paired t-test showed no statistical difference between positive derivatives. A p value of 0.02 was found between the negative derivatives of both groups.

:. . . . . . . .

i ! i i ........

i ........

V

i

~21~ so i....... i........i........ i i.............. i i i........i........:.i....... ii iiiiii) i.......i

.io ,.io f

i.jo i.o f (Hz)

,,~,

Fig. 5B. Power spectrum of left ventricular pressure obtained from Sentron catheter.

125 -4- 13 mmHg for the Millar and 125 & 13 m H g for the Sentron. Figure 2 (same data points on Fig. 2a and Fig. 2b) shows a comparison of end-diastolic pressure; range of EDP was: 2.2 to 20.7 mmHg (mean: 9.6 4- 5.2 mmHg) for the Millar catheter and 6.1 to 14.6 mmHg (mean: 10.3 4- 2.7 mmHg) for the Sentron catheter. Linear correlation coefficients are shown in Table 1. The correlation coefficient for the linear curve fit was 0.809. A quadratric curve fit was also calculated for the end-diastolic pressure (Fig. 2b and Table 1). The correlation coefficient was 0.916 for this curve fit. A paired t-test showed no significant differences in systolic and end-diastolic pressures between both groups.

Derivatives The first derivatives of the 42 pressure episodes were digitally calculated. A comparison of the maximal positive derivative (dp/dt)+ and of the negative (dp/dt)- are shown in Fig. 3 and Fig. 4, respectively. Positive derivatives had a range between 1204 and 3703 mmHg/s (mean: 2010 4- 663 mmHg/s) on the Millar and of 1140 to 3554 mmHg/s (mean: !959 i 690 mmHg) on the Sentron catheter. The negative derivatives had a range between -3276 and -2088 mmHg/s (mean: -2614 -t- 300 mmHg/s) on the Millar and of-4117 to -2063 mmHg/s (mean: -2838 4526 mmHg/s) on the Sentron catheter. Correlation coefficients for a linear curve fit are also shown in

Cross correlation The normalized cross correlation function was computed for all 42 episodes. The mean value for the maximal peak was 0.61 4- 0.04 with a 95% confidence interval of 0.0109 and 99% confidence interval of 0.0143. The range was 0.56 to 0.70. The timing difference was always equal or smaller than 1 ms, which is the minimal time resolution.

Frequency content Typical power spectra obtained from the Millar and Sentron transducers are shown respectively in Fig. 5a and Fig. 5b. The location of the peaks was measured on spectra from both transducers. A paired t-test showed no statistical difference between the location of the peaks on frequency spectra of both pressure sensors.

Discussion The study of pulsatile pressure in the heart and the circulation requires accurate measurements with available manometers. Most often fluid filled catheters are used with external placement of the transducer. The principal reason is its ease of use and low cost, ability to 'zero' without catheter withdrawal, connectable to a wide variety of catheters (diameters, m a t e r i a l . . . ) . However, the drawback is a less faithful reproduction of the pressure events due to inadequate damping characteristics of the setup, which corresponds to a second order system [8]. A serious obstacle to achieving even the theoretical limited response consists of the presence of occult air bubbles that may lodge anywhere in the catheter, stop cocks, or manometer. Therefore for high fidelity measurements, only tip catheters should be used. In this study a Millar microtip catheter was used as a 'golden standard', because of its known wide frequency bandwidth (viz. high dynamic accuracy). It is more or less considered as an industry standard but it is relatively costly. Both catheters were used simultaneously in the left ventricle. It was made sure (under fluoroscopic control) that both sensors were not in mechanical contact and at a distance of

82 approximately 2 cm. At this distance it is unlikely that there is electrical interference: the distance between transducers on catheters with multiple sensors is of the same order. Therefore electrical fields induced by currents in the sensors must be negligible at that distance. Pressures The Sentron tip catheter satisfies all requirements for a high fidelity manometer: an xy-plot of both tracings shows values around the identity line in the pressure range from 0 to 110 mmHg. Also systolic maxima (Fig. 1) compare very well. There is less correlation between end-diastolic pressures. There are four possible reasons for this finding: - The mechanical calibration in a pressure chamber was optimized for the 100 to 200 mmHg range. Mechanical calibration at 10 or 20 mmHg could enhance the accuracy in this range. has to be indicated manually on the pressure tracings. This can lead to some variability in the values.

-EDP

pressure differences between sensors not in the same horizontal plane.

-Hydrostatic

-

Physiologic pressure gradients within the ventricle during diastole [9].

On the other hand data points on Fig. 2a suggests a non-linear behaviour: indeed a quadratic curve fit shows a better correlation (Table 1). However, due to the physical characteristics of the piezo resistive material, it is unlikely to be caused by the transducer: a non-linearity would rather occur at high pressures. Differences between the EDP measured by Sentron and by Millar indicate that 7 outlier points are responsible for this behaviour. Therefore the most probable reason remains a lack in accuracy of the mechanical calibration at these pressure levels. Derivatives There is a better correlation between the peak positive derivatives (Fig. 3) than between the negative derivatives (Fig. 4) of the pressure tracings. The latter even showing a statistical significant difference. Physiology commands a different behaviour to the pressure-time curve during isovolumic contraction than during isovolumic relaxation. During the former period there is a continuous transition from a concave to a convex curve, as shown by the behaviour of the second derivative, whereas during the latter period a singularity is present at the closure of the aortic valve: a transition from a

polynomial decrease to an exponential decrease. Due to the numerical differentiation procedure the latter behaviour will give rise to a larger scatter in values. The Sentron system has a frequency bandwidth of 0 to 180 Hz, whereas the Millar from 0 to 20000 Hz. However, 20 harmonics are sufficient for obtaining derivatives [10]. During these expriments the highest heart rate was 90 beats/min. Therefore in this case a bandwidth of 30 Hz would already be sufficient for adequate derivative values. Even in case of a high rate of 180 beats/min a bandwidth of 60 Hz would be sufficient. Also as indicated on Fig. 4 and determined by differences between Sentron and Millar values, only 5 outlier points are largely responsible for the scatter of the negative derivative values. Differences in offset do not influence dp/dt. Differences in gain (sensitivity) would yield a systematic deviation. It was tacitly assumed that the signal on both transducers was exactly identical. Since the transducers are approximately 2 cm apart in the mongrel dog heart, pressure gradients in the heart [9] may be a (or the only) source of variation. This was not verified further. Cross correlation The cross correlation function describes the dependence of one waveform with another. The peak value of the normalized correlation function will be unity when the two functions are at a relative displacement that makes them identical over the period of observation. The mean value of the maximum of the normalized cross correlation function between both pressure tracings of 0.61 describes a close correlation. However, some parts of the pressure tracings are dissimilar. This finding was also observed for discrete points (maximal systolic pressure and end-diastolic pressure, Table 1) were the coefficient of the linear regression is different from unity. On the other hand no phase lag between both pressure tracings was observed at the time resolution of 1 ms. Frequency content The frequency response of the Sentron: 0 to 180 Hz as given by the manufacturers, is adequate for the power spectrum of the pressure waves (Fig. 5a and b) and for derivatives as well. This frequency range might be insufficient if an intra cardiac phonocardiogram is

83 needed [11]. Peaks in the power spectra are coincident. The resistance of the transducer to defibrillation currents has not been tested. The tip as well as the external amplifier should be protected against high currents. The long term stability (2 to 3 hours) shows a maximal total drift of 3 mmHg, period after which the sensor remains stable. This catheter was not conceived for chronic implantation. However, over a period limited to several days no influence on the performance is expected. Resterilization (with ethylene oxide gas) has no influence on the performance of the transducer. As a conclusion it can be stated that although the quest for an ideal unit remains, this device presents a valuable aiternative to the fluid filled systems with external pressure transducer and compares to existing devices, but at a lower price. The frequency response is more limited, preventing its use for intracardiac phonocardiography. The sensitivity is also less than of the Millar catheter. However, these days also economic arguments are becoming important in the decision making process in the medical community. With a price ratio between a fluid filled catheter, a Sentron and a Millar estimated as: 1:10:30, this motive can also come into play. In view of these characteristics and with its limitations, the Sentron catheter-manometer offers an economic solution when accurate pressure measurements are needed.

Acknowledgements The authors gratefully acknowledge the technical assistance of M.T. Stassen. We thank Dr.Ir. E. Ligtenberg and Mr. H. Kerkhoven for their critical review of the manuscript.

References 1. Marey EJ. Physiologie m6dicale de la circulation du sang. A. Delaheye 6dit., Paris, 1863. 2. Gmnbanm OE On a new method of recording alterations of pressure. J Physiol 1897; 22: 49-51. 3. Laurens P, Bouchard E Comu C, Baculard R Souli6 R Bruits et pressions cardiovasculaires enregistrds in situ ~t l'aide d'un micromanom~tre. Arch Mal Coeur 1959; 52: 121-9. 4. Millar HD, Baker LE. A stable ultraminiature catheter-tip pressure transducer. Med Biol Eng 1973; 11: 86-9. 5. Gould K, Treuholme S, Kennedy J. In vivo comparison of catheter manometer systems with the catheter-tip micromanometer. J Appl Physiol 1973; 34: 263-7. 6. National Academy of Sciences: Guide for the care and use of laboratory animals, NIH # 80-23, 1978 7. Bendat JS, Piersol AG. Random data: analysis and measurements procedures. J. Wiley Publ., New York 197 l: 28-31. 8. Geddes LA. Fundamentals of blood pressure transducers. In: Indwelling and implantable pressure transducers. CRC Press, Cleveland, Ohio 1977: 5-9. 9. Courtois M, Kovacs SJ, Ludbrnok PA. The transmitral pressure-flow velocity relationship: The importance of regional pressure gradients in the left ventricle. Circulation 1988; 78: 661-71. 10. Hellige G. Recording of ventricular pressure by conventional catheter manometer systems. I. Minimal requirements of blood pressure recording systems and estimation of frequency response characteristics. Basic Res Cardiol 1975; 71: 319-36. 11. Aubert AE, Denys BG, Meno E Reddy PS. Investigation of gallop sounds in dogs by quantitative phonocardiography and digital frequency analysis. Circulation 1985; 71: 987-93.

Address for correspondence: A.E. Aubert, Department of Cardiology, University Hospital Gasthuisberg, Catholic University Leuven, Belgium