Right Ventricular Volumetry by CatheterMeasurement of Conductance

JOHN C. WOODARD, CHRISTOPHER D. BERTRAM, and BARRY S. GOW*

From Centre for Biomedical Engineering, University of New South Wales, Kensington. Australia,and "Department of Physiology, University of Sydney, Camperdown, Australia

WOODARD, J.C. ET AL.: Right ventricular volumetry by catheter measurement of conductance. Theelectrical conductivity of hlood is sufficiently higher than that of myocardium to make feasible the detec-tion of cardiac volume changes hy measurement o/intraventricuiar jluid conductance. An eight-electrodecatheter was used to inject an aiternating current (100 ̂ lA or less, at 1500 Hz) via the two electrodes nearestthe ventricular base and apex, then the resulting five voltage differences between adjacent pairs of the sixintervening electrodes were measured. When current amplitude was held constant, the cross-sectional areaof the ventricuiar cavity slice defined hy planes perpendicular to the catheter through the relevant pair ofelectrodes was inversely proportional (to the first order) to the voltage di^erence. Measurement of multiplesegments compensated for isovolumic cavity shape changes. The technique had previously been shown tomeasure left ventricular volume successfully, but the geometry of the right ventricle made this measure-ment more problematical. Using open-chested, anesthetized greyhounds, we compared the catheter-mea-sured right ventricuiar volume change with stroke volume as measured by a pulmonary arterial electro-magnetic hlood flowmeter. With optimal catheter placement, good correlation between stroke volume andcatheter-measured volume changes was achieved when stroke volume was perturbed on a heat-to-heathasis. In six data records from three dogs, involving two different means of varying stroke volume (rapidinjection of blood and sinus node irritation), the correlations yielded r̂ values between 0.82 and 0.98. Themethod detected ineffective fnone/ecting) beats associated with normal-appearing QRS complexes and wasthus a more reliable indicator of cardiac mechanical Junction than an eiectrocardiogram. {PACE, Vol. 10,July, 1987)

voiumetry, catheter, impedance, conductance

Introduction

Measurement of cardiac volume by means ofa conductance catheter has been mainly investi-gated thus far in the left ventricle.^'^ While resultsfrom these investigations have yielded good cor-relations in animal studies, for either acute orchronic human applications, it would be prefera-ble to catheterize the right ventricle for reasons ofsafety and convenience. Possible application ofthis technique as a sensor for rate-responsive

Address for reprints: Dr. Christopher D. Bertram, Centre forBiomedical Engineering, University of New South Wales, P.O.Box 1, Kensington, Australia, 2033.

Received May 20, 1986; revision received August 13, 1986;accepted August 14, 1986.

pacing and implantable defibrillators would alsodictate a right heart location, particularly if itwere possible to locate the conductance-measur-ing electrodes on the pacing catheter.

This paper primarily reports results of rightheart volumetry obtained from anesthetizedopen-chested greyhounds, but it also includesmeasurements from experiments in the left ven-tricle designed to verify our method against re-sults in the literature.

Methods

The in-vivo experiments were conducted onanesthetized, open-chested greyhounds and thein-vitro work was done using excised greyhoundhearts. The conductance-measuring catheter

862 July-August 1987, Parti PACE, Vol. 10

RIGHT VENTRICULAR VOLUMETRY

Catheter

Alternating CurrentSource

(three other channels not shown)

J—

I

Sum

Differential Fi l ter DemodulatorAmplifier

Figure 1. Schematic diagram of right heart volumetry apparatus. The eight-electrode catheter isshown positioned in ihe right ventricle via the tricuspid valve. Numbers on caiheter electrodescorrespond to connections to electronics. Only two volume channels (of jivej are shown.

volumeOut

used in all investigations consisted of eight plati-num ring electrodes spaced at intervals of 10 mmalong a 7F shaft.* It additionally had a centrallumen with a hole at the distal end which allowedmeasurement of ventricular pressure and thusaided in initial positioning of the catheter.

Figure 1 illustrates the connections to theelectrodes and the usual position of the catheterin the heart. Electrodes 1 and 8 (the outermostelectrodes) were used to supply an amplitude-reg-ulated current of 100 nA at a frequency of 1500Hz. The differential voltages measured betweenpairs of the six remaining electrodes are inverselyproportional to the conductance of the bloodwithin the five ventricular segments containedbetween these electrodes. The boundary of thesesegments was defined by the equipotential sur-faces passing through the catheter electrodes andthe ventricular wall. This assumes that the con-ductance of the ventricular wall is negligiblecompared with the conductance of the blood.

The choice of 1500 Hz for these experimentswas made initially as a compromise between alow frequency which would ensure a high com-mon-mode rejection ratio in the amplifiers and ahigh frequency which would afford good attenua-

tion of the ECG (artifact) with active filters cen-tered on the catheter excitation frequency. Insubsequent measurements of the conductivity ofblood and myocardium, it was found that theratio of these two conductivities at 1500 Hz washigher than at the frequency used by Baan et al.̂of 20 kHz, confirming a finding by McKay et al.^This larger ratio makes the lower frequency moresuitable, as the conductance of the myocardium isone of the factors limiting the accuracy of thismethod.^

Validation in the Left Ventricle

Initial experiments were performed in theleft ventricle to validate our method against thework of others. The eight-electrode catheter wasintroduced into the left ventricle by cannulationof the right carotid artery and was positioned byfluoroscopy so that the distal end lay at the apexof the ventricle. During positioning of the catheterand during volume measurement, ventricularpressure was measured via the lumen in the con-ductance catheter using a Sanborn** 267B pres-sure transducer. Simultaneous pressure and con-ductance signals were digitized at 200 Hz using

• Webster Laboratories, Altadena, CA, USA.** Sanborn: now represented by Hewlett Packard, Palo Alto,CA, USA,

PACE, Vol. 10 July-August 1987, Part I 863

WOODARD, ET AL.

a PDP 11/03 computer with a 12-bit A/D con-verter. These signals were later calibrated and theconductance values were converted to volumeusing the equation given by Baan et al.'':

v(t) = - 2 Ci

where V(t) is the volume as a function of time,L is the distance between voltage

sensing electrodes,is the specific conductivity of

hlood,and C is the measured conductance of

each segment calculated as theratio of current to voltage be-tween tbe two electrodes defin-ing the segment.

In-Vitro Experiments in the Right Ventricle

In order to determine whether an eight-elec-trode catheter (i.e., five volume segments) wassufficient to eliminate the effects of isovolumicventricular shape changes in the right ventricle, aseries of experiments was performed in excisedhearts.

The eight-electrode catheter was insertedinto the right ventricle by cannulation of a pulmo-nary lobar artery. The right ventricle was sealedat the pulmonary artery around the catheter andat the tricuspid valve by suturing and gluing thesides of the valve ring together. Having evacuatedthe ventricular cavity by syringe via the lumen ofthe conductance catheter, a measured volume ofblood was then introduced into the ventricle viathe same lumen.

By gentle squeezing of the external surface ofthe right ventricle between two fingers, it wasthen possible to move its contents thus simulatingthe effect of an in-vivo isovolumic shape change.Conductance signals from the catheter were re-corded during these manipulations and the re-sulting segmental volumes were calculated as afunction of time.

It was also possible to simulate the effect ofmeasuring only one large volume segment en-compassing the entire ventricle as if using onlytwo voltage-sensing electrodes at opposite ends ofthe ventricle. This was achieved by adding thefive segmental voltage differences to yield the

overall ventricular conductance and then con-verting this to volume. Because of the nonpris-matic shape of this large single segment, it wasexpected that isovolumic shape changes might beerroneously transduced as apparent changes inventricular volume.

In-Vivo Experiments in the Right Ventricle

For a pacemaker application of this tech-nique, it is desirable to measure stroke volume ona beat-to-beat basis so that pacing rate could beadjusted rapidly to changed metabolic demandsas sensed by transient increments in stroke vol-ume. Correlation of this method against a stan-dard method, therefore, should also be done on abeat-to-beat basis and was done using an electro-magnetic flowmeter cuff, implanted on the pul-monary artery. By this means, instantaneousflow-rate from the right ventricle could be mea-sured and by integration, systolic volume changesin the right ventricle could be computed. Theflow-rate signal was digitally integrated over theperiod of each cardiac cycle and then reset to zeroat the beginning of the next cycle. The time forreset of the ejected volume to zero was chosen tobe coincident with the R peak of the electrocar-diogram.

In order to vary the stroke volume for thepurposes of producing correlations, two interven-tions were applied to the dog preparation. Thefirst method, exploiting the Frank-Starling mech-anism, was to increase venous return by rapidinjection of a bolus of previously withdrawnblood into a jugular vein. The other method wasinduction of extrasystoles by irritation of the rightatrial area near the sinus node. Those extrasys-toles which occurred soon after normal heartbeats caused smaller volumes of blood to beejected than with a normal beat due to the lack offilling time. The subsequent correspondingly ele-vated beat was also abnormal as a result of in-creased filling time while the sinus node recov-ered its normal rhythm.

During the right heart experiments the fol-lowing signals were recorded: pulmonary arterialflow-rate, the five catheter voltage outputs, ECGand arterial pressure. Systemic arterial pressurewas measured using a Sanborn P267B transducer

864 July-August 1987, Part I PACE, Voi, 10

RIGHT VENTRICULAR VOLUMETRY

202^^O)X

EE

LUCC

COCOLUCCQ.

251 10 VOLUME (ml) 140

Figure 2. Left ventricular pressure vs left ventricular volume recorded from impedance catheterinserted via mitral valve. Four phases of ventricular activity can be seen, clockwise from bottom:filling, isovolumic contraction, ejection, relaxation.

attached to a fluid-filled femoral catheter withthe distal end in the abdominal aorta.

Results

Left Ventricle In-Vivo

Pressure-volume plots obtained using theeight-electrode catheter are seen in Figure 2. Fig-ure 3 gives the individual segmental volumesplotted against the ventricular pressure. SegmentA is nearest the ventricular apex and segment E isnearest the aortic valve.

Right Ventricle In-Vitro

The effect of manual squeezing of the heartnear the ventricular apex on segmental volumesis seen in Figure 4, the squeeze having occurred0,9 s from the start of the record. The total volumeof the ventricle, i.e,, the sum of these five individ-ual volumes, is given in Figure 5 as is the effect ofusing only two sensing electrodes (as simulated bythe method outlined).

Right Ventricle In-Vivo

Rapid injection (about 1.5 s) of 58 ml of bloodinto the jugular vein 3 s after the start of therecord caused increased stroke volume as shownby the flow-rate signal in Figure 6. The total vol-ume ejected as computed by integration of theflow signal is plotted on the second axis. Rightventricular volume measured by the catheter isgiven on the first axis.

Ventricular volume computed from the cath-eter is plotted in Figure 7 against ventricular vol-ume derived by integrating the electromagneticfiow signal. A line of best fit is shown which had acorrelation coefficient (r̂ ) of 0.86. Also shown isthe line of identity (dashed line).

Figure 8 shows the result of irritation of theatrial region near the sinus node. An extrasystoleoccurred at approximately 1.9 s and at 3.5 s afterthe beginning of the recording. These extrasys-toles resulted in very small ejections of blood asevidenced from the electromagnetic flow trace on

FACE, Vol. 10 July-August 1987, Part I 865

WOODARD, ET AL.

(A)

(B)

VOLUME (ml)

VOLUME (ml)

Figure 3. Left ventricuJar pressure vs individual voiume segments. Graph A is from the voJumesegment nearest fhe apex of the ventricJe; graph E is from the segment nearest the base, probablyextending beyond the mitral valve into the left atrium.

the fourth axis. Following these extrasystoles, asmall compensatory pause occurred and thennormal rhythm resumed with a very large strokevolume from the next beat. As with the previousfigure, pulmonary flow-rate signal and its integralappear on axes 2 and 4, respectively. The correla-tion coefficient of volume measured by the catb-eter vs tbe integrated flow-rate was 0.98.

Overall correlations from six data recordsobtained from tbree dogs varied from 0.82 to 0.98for the six data records. Lines of best fit typicallyexhibited slopes of approximately 0.5.

Discussion and Conclusions

Plots of pressure vs total volume (transducedby tbe catbeter) for tbe left ventricle sbow tbeexpected, approximately rectangular format. In-dividual volume segments defined by equipoten-tial surfaces through the catheter electrodes alsoshow a rectangular format except segment E inFigure 3. This segment either lay beyond the mi-tral valve or traversed it and made little contri-bution to ventricular ejection. The majority ofejected volume came from segments B, C, and D

866 July-August 1987, Part I PACE. Vol. 10

2.4

VOL (ml)1.54.1

VOL (ml)3-27.1

VOL (ml)6.37.8

VOL (ml)7

5.4

VOL (ml)4.6

APEX

KJf^f^f—^ tA ^ f ••— "" '"*"'

r\ BASE

2 5

O>

TIME (sec)Figure 4. In-vitro record from the right ventricle of an excised non-beating hearl; five individualsegments are shoK'n. At approximately 0.9 s, the righl ventricle was externally compressed nearthe apex, forcing hlood from apex to hase. Time scale = 0.5 s/division.

5 segments

1 segment

TIME (sec)Figure 5. Tola! volume of the ventricle obtained from the previous experiment. The sum of thelive volume segment records from the previous figure appears on the top trace ivith the simulatedsingle-segment volume on the lower trace. Negligible change in totaJ volume is observed duringisovolumic shape changes when the ventricular cavity volume is measured using jive segments,but a large arti/actual change in volume is observed if a single large segment is used. Time scale= 0.5 s/division.

PACE, Vol. 10 July-August 1987, Part I 867

WOODARD. ET AL.

5 4CATHETER

VOLUME (ml)

3428

INTEGRATEDFLOW-RATE

(ml)01

FLOW-RATE(l/min)

TIME (sec)Figure 6. Catheter-derived right ventricular volume plotted with /low-rate signal /rom an electro-magnetic cuff on the pulmonary artery. Centre trace is the digitally integrated/low-rate signal,reset to zero synchronously with the QRS complex of the ECG. Changes in stroke volume werebrought ahout hy a rapid injection of hlood into the jugular vein. Time scale = 0.8 s/division.

in Figure 3, these three contributing 81% of thetotal stroke volume.

Calculation of segmental volume from theconductance of the segment also assumes thateach segmental volume as defined ahove is pris-matic in shape, i.e., it has constant cross-sectionalarea hetween the bounding equipotentials. If theaxial lengtji of the segments is small compared toa notional radius calculated from their cross-sec-tional area, this assumption is fairly well satisfied.For either the right or left ventricle, which ex-hibit a tapering shape, it is necessary to use anumber of segments in order to satisfy the re-quirement for prismatic volume shape. The use ofan eight-electrode catheter effectively divides theventricular cavity into five volume segments,each of which approximates a prismatic shape.The practical consequence is that total volumewill be accurately transduced despite large varia-tions in the cross-sectional area along the axis ofthe ventricular cavity.

Localized changes in the ventricular cross-sectional area resulting in movement of bloodwithin the ventricle without a total ventricular

volume change are likely to occur during the iso-volumic contraction phase of the cardiac cycle,due to the differences in the rates of spread andmagnitudes of the contractions of the thick intra-ventricular septum and the thinner right ventric-ular free wall, if the multisegmental approach isvalid, such isovolumic changes should not betransduced by the catheter as a total volumechange when all segmental volumes are added.

Results from the in-vitro experiments in theright ventricle illustrate well the necessity for amultiple-segment approach to the measurementof volume in a highly irregular space such as theright ventricular cavity. Isovolumic shapechanges brought about by external compressionof the ventricular cavity were well transduced bythe five-segment method as shown in Figure 4 inwhich blood moved from the region near the apex(two top traces) toward the base. When these seg-mental volumes were added, the net change wasnegligible as would be expected since the totalventricular volume was fixed (Fig. 5). However, ifonly one large segment was used, i.e., using onlytwo voltage-sensing electrodes, one at either end

868 July-August 1987, Part I PACE, Vol. 10

(|ui)

ii

o

\TH

ETE

P

u

1 1

18

g

8

7

6

5

4

3

2(P

'CD

J__j I

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

INTEGRATED FLOW-RATE (ml)

Figure 7. Cafheter-derived volume changes plotted against ejected volume computed from theintegral o/pulmonary arterial /low-rate. Data from Figure 6.

2 2 2rCATHETER

VOLUME (ml)179

27INTEGRATEDFLOW-RATE

(ml) -2

ECG

9.8FLOW-RATE

(l/min)- 5

- A

TIME (sec)Figure 8. Similar pJot to that of Figure 6 but incorporating ECG on the third trace. Changes instroke volume were brought about by sinus node irritation, resulting in extrasystoJes and compen-satory pauses in cardiac rhythm. Time scaJe = Q.Q s/division.

PACE. Vol. 10 July-August 1987, Part I B69

WOODARD, ET AL.

of the ventricle, a large artifactual change in totalvolunie was transduced when the containedblood was moved within the ventricle by externalsqueezing as shown in the lower trace on Figure5. A reduction in the apparent volume sensed bythe catheter also occurred with a large nonpris-matic volume segment. This was due to the domi-nation of the measured conductance by any nar-row regions within that segment.

Data from the right ventricle gave good cor-relations between ejected volume (as measuredby the electromagnetic flowmeter) and changes inright ventricular volume (as measured by thecatheter) for both interventions.

In all catheter-derived measurements of ven-tricular volume, an overestimate of the true vol-ume occurs. This is thought to be partly due to theshunting of catheter current through the myocar-dial wall and the blood within the left ventricle.This shunting of current would create an increasein apparent conductance observed between thecatheter electrodes and thus mimic the additionalvolume that was observed.^

Changes in right ventricular volume, how-ever, appear to be underestimated by tbe cathetermethod. One factor that contributes to this is in-complete transduction of the total ventricularcavity by the catheter. The anatomy of the rightventricle is such that the tricuspid valve (theentry point of the catheter) is approximately half-References

1. Baan J, van der Velde ET, de Bruin HG, et al. Con-tinuous measurement of left ventricular volume ofanimals and humans by conductance catheter. Cir-culation 1984; 70:812-822.

2. McKay RG, Speers JR, Aroesty JM, et al. Instanta-neous measurement of left and right ventricularstroke volume and pressure-volume relationshipswith an impedance catheter. Circulation 1984;69:703-710,

3. Salo RW, Wallner, TG. Computer modelling of in-tracardiac impedance plethysmography. I.E.E.E./

way between the ventricular apex and the pulmo-nary valve. Thus, the catheter electrodes traverseapproximately half of the right ventricular vol-ume. Changes in ventricular volume occurringbetween the tricuspid valve and the pulmonaryvalve which contribute to ejected volume will notbe expected to appear as a portion of the cathetersignal. Undoubtedly other effects also contributeto the catheter output. The known flow-depen-dence of the electrical conductivity of blood,^ as-sociated with the generation of net "preferred"orientation of red cells in the presence of a veloc-ity gradient, could give rise to a phasically chang-ing component of catheter output that also variedwith catheter position within the ventricle.

In conclusion, the data presented here showclearly that the conductance catheter method cangive a reliable measure of stroke volume changesfrom the right ventricle. Various factors combineto render impractical the absolute calibration ofventricular volume via the theoretical governingequation and knowledge of the blood conductiv-ity, but this may not be critical in pacemaker ap-plications. Our subsequent work is being directedtowards the strict definition of the conditionswhich must be satisfied for successful use of themethod in the right ventricle, and in assessinghow restrictive these conditions are in the con-text of the applications visualized.

N.S.F. Symposium on Biosensors. Los Angeles, Cali-fornia, 1984; 29-31.Baan J, Jong TTA, Kerkhof PLM, et al. Continuousstroke volume and cardiac output from intraven-tricular dimensions obtained with impedancecatheter. Cardiovasc Res 1981; 15:328-334.Visser KR. Lamberts R, Korsten HHM, et al. Obser-vations on blood flow-related electrical impedancechanges in rigid tubes. Pflugers Arch 1976; 366:289-291.

870 July-August 1987, Part I PACE, VoL 10