Measurement 24 (1998) 207–215

Design and development of a new ultrasonic doppler techniquefor estimation of the aggregation of red blood cells

*E. Karabetsos , C. Papaodysseus, D. KoutsourisBiomedical Engineering Laboratory, Department of Electrical and Computers Engineering, National Technical University of Athens,

20 Anastasiou Gennadiou Str., 11474 Athens, Greece

Received 24 June 1997; received in revised form 1 July 1998; accepted 10 August 1998

Abstract

Aggregation of red blood cells (RBCs) is one of the principal hemorheological factors which plays an important role incapillary circulation. In order to study the RBC’s aggregation, an ultrasound Doppler in-vitro technique, using pulsed wavemonoelement pencil 4 MHz probes, has been designed. An hydraulic pump system has been implemented, establishing alaminar blood flow profile into a rectangular cross-section plexiglass tube. Adding dextrans to blood samples, red blood cellsaggregation has been achieved and observed for various hematocrit values. Both the emitted and the backscattered signals,were driven to a system containing a multi-channel digital oscilloscope – of high sampling rate and processing capabilities –and a powerful PC with a high acquisition A/D card and special control software. Quantitative detection and measurementof the aggregation phenomenon can be achieved according to the experimental conditions and parameters used. 1998Elsevier Science Ltd. All rights reserved.

Keywords: Ultrasonic technique; Pulsed Doppler; Erythrocyte aggregation

1. Introduction Several techniques have been used to study aggre-gation phenomenon: measurement of erythrocyte

Red Blood Cells’ (RBCs) aggregation has been sedimentation rate [2], viscometry [1], microscopicwidely studied and its importance is well established and microphotographic observation, photographicalin the rheology of microcirculation [1–3]. It is a stereological methods [5,6], TV and computerizedmajor factor responsible for the flow properties of image analysis [7], light reflectometry [8], lightblood and therefore the measurement of red blood transmission [9], optical scattering [10–12] andcells’ (or erythrocyte) aggregation is rheologically ultrasonic backscattering [13–19].important for quantifying flow abnormality in All these methods differ not only in their principlepathological conditions [1]. The rate and degree of but also in the significance of their results. Each oferythrocyte aggregation depend on the physico- these techniques has contributed significally to thechemical properties of the erythrocytes, the suspend- quantification of erythrocyte aggregation, but most ofing medium and the presence of macromolecules and these methods can provide only an indirect means ofthe flow parameters and conditions [4]. quantifying erythrocyte aggregation using an index

and none of them can be applied in-vivo.*Corresponding author. E-mail: thkara@softlab.ntua.gr Ultrasonic methods can be utilized in-vitro and

0263-2241/98/$ – see front matter 1998 Elsevier Science Ltd. All rights reserved.PI I : S0263-2241( 98 )00053-0

208 E. Karabetsos et al. / Measurement 24 (1998) 207 –215

in-vivo as well, to get information on the aggregate 2. Theoretical aspectsstructure of red blood cells. In-vitro experiments onacoustical detection of erythrocyte aggregation in The Doppler signal is a composite of echoes fromblood suspensions, in flow or stasis, have been a large number of red blood cells. Even if these cellsreported by many research teams. Some of them move with identical velocities, the composite signalobtained in-vitro ultrasonic images of whole blood will not be a single sinewave because particles (cells)suspensions, showing the aggregation phenomenon continually enter and leave the sample volume atand measures of blood echogenicity with flow pa- random times. So, a Doppler signal is not a simplerameters – velocity and shear stress – hematocrit, amplitude and frequency modulated signal in whichetc., using common commercial echographic systems the frequency modulation follows the scatterer’salong with especially designed measurement apparat- velocity. It includes random characteristics due to theuses [15,16,18,19]. Others, [13,14,17] use ultrasound random phases of scattering particles present in theA-mode probes and measurement cells in in-vitro sample volume. Other effects such as geometricexperiments, measuring the intensity of the back- broadening and spatially varying velocity also affectscattered signals from erythrocytes suspensions and the signal [26].the ultrasound backscattering coefficient, trying to The following equation:evaluate aggregation parameters, such as the aggre-

t rˆS (t) 5 C D cos(vt 1 w )u (t)u (t) (2.1)j j j j j jgates’ size, and to examine the effects of manyparameters – hematocrit, dextran concentration, red

describes the scattered signal from a single cellblood cells’ electrical charge and shear stress – onwithin the sample volume as a function of time t,erythrocyte aggregation.

ˆwhere v5v 1v , v is the carrier frequency and vo d o dThe Doppler ultrasound technique is now com-is the doppler frequency and the phase w is randomjmonly used in noninvasive blood flow examinationand depends on the particle position. The subscript j[20,21]. Doppler frequency spectrum is derived fromdenotes the particle number. C is the amplitude ofjechoes backscattered by red blood cells in thethe two-way beam pattern at the location of particleflowing blood suspensions. There is abundant theo-j. D is the scattering coefficient for the particle. Thejretical, but rather little experimental work, conducted t rtwo functions u (t) and u (t) are, respectively, thej jby some research teams for the modeling of theenvelope of the transmitted pulse and the shape ofbackscattered Doppler ultrasound signal from bloodthe receiver gate. It is assumed that the particle[22–24] and for the modeling of appropriate bloodvelocities are much less than the sound speed, so thatflow systems for Doppler ultrasound studies [25].cell locations do not change within a single pulse. ItPulse-wave Doppler technique which is more widelyis also assumed that the direction of the waveused at the present time, has the decisive advantagenumber vector, and hence theta does not varyof providing blood flow information at a specifiedthrough the sample volume and that the diameter ofdepth on the ultrasound beam axis [20,21], hencethe vessel is large compared to the size of the sampleallowing very specific and accurate observations involume.the red blood cell suspension.

The entire sample volume, after demodulation is aIn this work, a hydrodynamically well-definedsum of cosines and hence has the form:in-vitro measurement device has been developed

using a pulse doppler experimental arrangement set- S(t) 5 K(t) cos(v t 1 w )d jup, integrated in a system containing a digitaloscilloscope and a powerful personal computer with Both the amplitude and the phase change wheneverspecial software connected via GPIB interface. the arrangement of particles within the sampleQuantitative detection and measurement of the volume changes. The particle arrangement changesaggregation phenomenon can be achieved accord- with each pulse because old particles exit the sampleing to the experimental conditions and parameters volume and new ones enter and because the am-used. plitude of the signal from each particle changes with

E. Karabetsos et al. / Measurement 24 (1998) 207 –215 209

location as the particle moves through the beam cm from the shaft. It has a length of 3.5 cm in thepattern. upper wall surface of the tube and 8 cm in the lower

one. This is the right distance from the motor inorder to be sure that laminar steady blood flow has

3. Material and method been established. This length L has been calculated,based on the annular velocity (2 r / s) of the arm, on

The experimental setup of the measurements is the maximum mean value of the fluid velocity in thebest described in the block diagram of Fig. 1. A motor end and in the entrance of the orthogonic tube.specially designed hydraulic pump system has been In the equation used for this calculation [20,24] theimplemented, using an electric step motor that length: L5k?d ?R , where k is a constant k50.04,h e

establishes a steady, laminar blood flow profile into a d is the hydraulic diameter of the rectangular tube-h

rectangular cross-section plexiglass tube. This step based on its’ dimensions x, y (x51.2 cm, y51.5 cm)motor is electronically controlled in order to main- and R is Reynolds number which equals 990 in thise

tain a constant number of rotations per second. The case, assuring thus the establishment of the steadysignal that conducts the step motor has a sawtooth laminar flow in the measurement window, with aprofile. The motor and the sampling are synchronised velocity of 25 cm/s. This window is surrounded byvia a specifically programmed microcontroller. The an external chamber, made also of plexiglass, with atube has a cross-section area of 1.2 cm (width)31.5 height of 15 cm and a length of 20 cm. It iscm (height), with a wall thickness of 4 mm. The step constantly full of distilled water and has two adjust-motor used is attached to a thrusting mechanical arm able supporting arms, where the ultrasonic probes areof 11 cm, with an eccentricity of 1.5 cm. The arm driven from. All details of this, specially designedleads to a shaft (1.2 cm diameter) attached to the and developed in-vitro measurement device arerectangular tube, which supplies our device with a described in Fig. 2.continuous step front–back movement, applied to the The ultrasonic probes used in the experimentalused blood samples. With this pump system, a setup are monoelement, pulsed-wave (PW) Dopplersimulation of the laminar blood flow profile and the ultrasound, pencil probes operating in the frequencyflow conditions that exist in the large arteries in of 4 MHz, with an adjustable focal zone of 10–50general, have been achieved. mm. The Doppler Ultrasound PW mode offers the

The measurement window of the device,made of advantage of easier acquisition of useful insonationelastic rubber membrane, is collated within the inner thanks to the controlled time receiving windows,surface of the 4 mm tube, in the exact distance of 55 permits large amplification within the useful time

windows and also allows the shifting of the receivingwindow inside the area of interest [20,21], enablingthus measurements at different depths. In the mea-surements held,the sample volume resided in thecenter of the tube lumen, with a constant angle ofinsonation equal to 538.

The block diagrams of both the emitter and thereceiver of the Doppler ultrasound device are shownin Fig. 3 and Fig. 4. The emitter circuit consistsmainly of a digital part, where the burst and samplesignals are created, and of an analog part, where thepulses that stimulate the probe and the demodulationsignal are created. External touch keys enable theadjustment of some circuit parameters, such as thenumber of transmitted pulses, the PRF, the depth of

Fig. 1. Block diagram of the experimental setup. the sample volume (residing within the focal zone of

210 E. Karabetsos et al. / Measurement 24 (1998) 207 –215

Fig. 2. In vitro measurement device.

Fig. 3. Doppler ultrasound pulsed wave Emitter.

the probes) and the sample pulse width. The burst repetition frequency (PRF) of 5 kHz (PRF range: 2and the sample signal (coming out from the m- to 17 kHz).The time-distance between the burst andcontroller) specify the time receiving windows, the sample signal specifies the exact depth of thewhere pulses stimulate the emitter probe at a pulse sample volume, which equals 2.5 cm in the in-vitro

E. Karabetsos et al. / Measurement 24 (1998) 207 –215 211

Fig. 4. Doppler ultrasound pulsed wave Receiver.

measurement device. At the receiver circuit, which is of various molecular weights (ranging from 37 700separately battery-powered (in order to minimize to 580 000).every noise coming from the emitter), the acquiredprobe signal, is firstly resistance coupled, linearlypre-amplified and then demodulated (the demodula- 4. Resultstion signal comes from the crystal oscillator of theemitter circuit). Then the signal is being low-pass With the presented method, qualitative and quan-filtered and passes through the sample and hold titative estimation of the aggregation phenomenonamplifier circuit. has been achieved. This was accomplished by defin-

The outcoming signal, passing through an instru- ing experimentally and theoretically the probe’smentation amplifier, is then driven to a digital sensitivity and resolution, a very crucial factor foroscilloscope (LeCroy model 9310A, with a band- estimating aggregates’ size. This is due to the factwidth of 300 MHz and a very high sampling rate). that the probe itself ‘‘sees’’ as minimum backscatter-An application software, Labview for Windows, and ing volume or elemental acoustic voxel than cana data acquisition card (DAC-GPIB-PCII) have been resolve, a volume imposed by its axial and lateralused for the remote control of the oscilloscope via a resolution directly related with the values of somepersonal computer. Thus the oscilloscope has been technical parameters described in the doppler elec-remotely controlled while receiving or sending sig- tronics section. So, the minimum size of detectablenals /messages (every function of it was simulated by aggregates is imposed by the (axial1lateral) res-its’ virtual GPIB instrument driver). olution of the used ultrasound probes. It is known

In the measurements held, the in-vitro device was that a theoretical elemental acoustic voxel withcontained in a temperature controlled water bath. dimension l /20 for 4 MHz contains around 60 cellsThe normal red blood cells’ suspensions have been [23], while the experimentally resolved volumeprepared from human blood collection bags. The inside the sample volume is filled with hundreds ofwhole blood sample was first centrifuged once in cells. The authors proceeded in a definition of the3000 r /min (for 5 min), then washed twice (also at minimum backscattering level received from the3000 r /min, for 5 min), using either serum w (0.9% probe as reference signal in the certain experimentalNaCl, 154 mM/l) or PBS buffer (NaCl 140 mM, conditions used, in order to set it as a minimum andKCl 2.7 mM, KH PO 1.5 mM, Na HPO 8.1 mM). quantify their results. This was done by taking2 4 2 4

After centrifugation and washing, the acquired RBCs signals from the in-vitro measurement device firstlyhave been suspended in serum w to create blood filled with serum w only and then adding gradually asuspensions of various hematocrits, of 220 ml (the very small amount of RBCs suspension and waitingtotal tube internal volume). Aggregation has been for it to settle and be completely diluted. Then fromachieved by adding Dextran (concentration 40 gr / l) the signals taken, the smaller value of the area

212 E. Karabetsos et al. / Measurement 24 (1998) 207 –215

parameter – as it will be discussed right below – wasmeasured and this was the relative minimum for allother area measurements, which had a multiple valueof the previous mentioned.

In Figs. 5 and 6 the amplitude value of eachacquired signal in relation to the consecutive samplepoints is drawn. Aggregates differencing in size andvelocity have been detected by their transit timethrough the window of observation and their relativebackscattered amplitude value levels. So, larger andslowly moving aggregates can be distinguishedbetween smaller and faster ones. Calibration of thewhole method and measurement device has been

Fig. 6. (a) Acquired signal from a suspension of hematocrit value(Hct) 40%. (b) Acquired signal from a suspension of hematocritvalue (Hct) 40% after adding a specific dextran (Dx 71).

performed with latex particles of known dimensions,close to these of a single red blood cell.

By adding dextrans of different molecular weightsand concentrations for a wide range of hematocritvalues, a large data base of backscattered ultrasounddoppler signals has been created. After the classifica-tion of these signals, various pattern recognitionmethods have been applied, mainly morphologicaland statistical ones. It results from the employedanalysis that the envelope,energy and area charac-teristics of the obtained backscattered signals areFig. 5. (a) Acquired signal from a suspension of hematocrit valuequite good indicators of the degree of RBCs aggrega-(Hct) 30%. (b) Acquired signal from a suspension of hematocrit

value (Hct) 30% after adding a specific dextran (Dx 580). tion. But the best method for the estimation of

E. Karabetsos et al. / Measurement 24 (1998) 207 –215 213

aggregates’ size is the calculation of the area of thedifferent parts within a signal, which is the depictionof consecutive sample points in time. This isachieved as shown in Fig. 7, by calculating the valueof the area included within the dc level and anycaptured amplitude values before they return to thedc level again, indicating the passage of a group ofscatterers. The calculation of each area value isperformed with the application of the Simpson’strapezoidal integration rule, on each acquired signal.

From the experiments held, the proposed methodseems suitable to detect and qualitatively describeaggregation phenomena for hematocrits up to 40%,while all other presented methods so far, work foressentially lower hematocrit values. So, by means ofthe proposed method a quantitative relation betweenthe presence of various dextrans and the degree of Fig. 8. Comparison of the aggregate sizes (computed with the areaaggregation has been established in-vitro. Also, the parameter and presented in groups) for various hematocrit values.distribution of the size of the formed aggregates inthe various experimental solutions has been deter-mined and an accurate relation between ultrasoundbackscattering and the actual size of the corre- depicted in Figs. 9 and 10. The aggregate sizessponding observed aggregates has been obtained. found, are shown in groups for better comparison

A comparison of the degree of aggregation be- and interpretation of results.tween employed suspensions with different values of The maximum aggregates’ sizes observed with thehematocrit is shown in Fig. 8. In addition, a com- presented method are found between hematocritparison of the degree of aggregation between a values of 20–30% and they peak for the hematocritsuspension of a specific hematocrit value and the value of 30%.same suspension after adding a certain dextran is

5. Discussion

It would be very interesting that more experimentswill be performed with this in-vitro apparatus withother substances that affect the mechanoelastic andaggregation properties of the red blood cells such asneuraminidase or glutaraldeyde in order to study theeffect of these solutions to RBC’s aggregation. Thiswill hopefully lead to whole-blood experiments,since measurement of aggregation kinetics in adynamic flow system is of major importance ingaining a proper understanding of the non-Newto-nian behavior of whole blood in large vessels.

A step further would be also to use an hydraulicpump system of smaller dimensions, in the range ofmm, to simulate real life blood flow conditions insmall arteries or even, with an appropriate hydraulicFig. 7. Calculation of the area parameter in some parts of a signal

from a suspension of hematocrit value (Hct) 20%. system, simulate bifurcated vessels.

214 E. Karabetsos et al. / Measurement 24 (1998) 207 –215

Fig. 9. Comparison of the results for the aggregate sizes, presented in groups, for a certain hematocrit value suspension (Hct530%) – H30,before and after adding a specific dextran (Dx 71) – D71H30.

Fig. 10. Comparison of the results for the aggregate sizes, presented in groups, for a certain hematocrit value suspension (Hct540%) – H40,before and after adding a specific dextran (Dx 580) – D58040.

E. Karabetsos et al. / Measurement 24 (1998) 207 –215 215

[12] E. Muralidharan, Simultaneous determination of hematocrit,It is of a great interest to raise the frequency of theaggregate size and sedimentation velocity by He–Ne laserultrasonic probes, getting to 8 MHz – another widelyscattering, Biorheology 31(5) (1994) 587–599.

used frequency value in diagnostic applications – in [13] M. Boynard, J.C. Lelievre, An approach of red blood cellorder to compare the ultrasound backscattering re- aggregation by ultrasonic echography, Clin. Hemorheol. 9sults of the formed aggregates. These results in both (1989) 771–779.

[14] M. Boynard, J.C. Lelievre, Size determination of red bloodfrequencies would strengthen the presented methodcell aggregates induced by dextran using ultrasound back-and so complete in-vitro modeling will have beenscattering phenomenon, Biorheology 27 (1990) 39–46.

achieved. The final aim is to perform in-vivo mea- [15] S.Y. Kim, I.F. Miller, B. Sigel, M.P. Consigny, J. Justin,surements with this technique in superficial vessels. Ultrasonic evaluation of erythrocyte aggregation dynamics,

Biorheology 26 (1989) 723–736.[16] C. Mahony, J. Ferguson, P.L.C. Fischer, Red cell aggregation

and the echogenicity of whole blood, Ultrasound Med. Biol.References18(6 /7) (1992) 579–586.

[17] S.M. Razavian, M.Th. Guillemin, R. Guillet, Y. Beuzard, M.[1] S. Chien, Biophysical behavior of red cells in suspensions. In Boynard, Assesment of red blood cell aggregation with

Red Blood Cell, Vol. 2, ed. D.M. Surgenor. Academic Press, dextran by ultrasonic interferometry, Biorheology 28 (1991)New York, 1975, pp. 1031–1133. 89–97.

[2] L. Dintenfass, Rheology of Blood in Diagnostic and Preven- [18] B. Sigel, J. Machi, J.C. Beitler, J.R. Justin, Red celltive Medicine. Butterworths, London, 1976, pp. 3–44. aggregation as a cause of blood-flow echogenicity, Radiolo-

[3] R. Fahraeus, The influence of the rouleau formation of the gy 148(3) (1983) 799–802.erythrocytes on the rheology of the blood. Acta Medica [19] B. Sigel, J. Machi, J.C. Beitler, J.R. Justin, J.C.U. Coelho,Scandinavica, 1958, CLXI, fasc. II, 151–165. Variable ultrasound echogenicity in flowing blood, Science

[4] J.F. Stoltz, M. Donner, Hemorheology: importance of eryth- 218 (1982) 1321–1323.rocyte aggregation, Clin. Hemorheol. 7 (1987) 15–23. [20] D.H. Evans, W.N. McDicken, R. Skidmore, J.P. Woodcock,

[5] L. Dintenfass, H. Jedrzejczuk, A. Willard, Photographic, Doppler Ultrasound-Physics, Instrumentation, and Clinicalstereological and statistical methods in evaluation of aggre- Applications. John Wiley and Sons, New York, 1989, pp.gation of red cells in disease. Part I: Kinetics of aggregation, 84–107.Biorheology 19 (1982) 567–577. [21] F.W. Kremkau, Doppler Ultrasound – Principles and Instru-

[6] H. Schmid-Schonbein, J. Gosen, L. Heinich, H.J. Klose, E. ments, 2nd edn. W.B. Saunders, New York, 1995, pp. 121–Volger, A counter rotating ‘‘rheoscope chamber’’ for the 175.study of the microrheology of blood cells aggregation by [22] L.Y.L. Mo, R.S.C. Cobbold, A non-stationary signal simula-microscopic observation and microphotometry, Microvasc. tion model for continuous wave and pulsed Doppler ultra-Res. 6 (1973) 366–376. sound, IEEE, Transactions on Ultrason, Ferroelec. Freq.

[7] S. Chen, G. Barshtein, B. Gavish, Y. Mahler, S. Yedgar, Control, UFFC-36, (1989) 522–530.Monitoring of red blood cell aggregability in a flow-chamber [23] L.Y.L. Mo, R.S.C. Cobbold, Theoretical models of ultrasonicby computerized image analysis, Clin. Hemorheol. 14(4) scattering in blood. In Ultrasonic Scattering in Biological(1994) 497–508. Tissues, eds. K.K. Shung, G.A. Thieme. CRC Press, New

[8] R. Brinkman, W.G. Zijlstra, N.J. Jansonius, Quantitative York, 1993, pp. 125–170.evaluation of the rate of rouleaux formation of erythrocytes, [24] K.K. Shung, G. Cloutier, C.C. Lim, The effects of hemato-studies with the aid of syllectometry, Proc. Kon. Ned. Akad. crit, shear rate and turbulence on ultrasonic doppler spectrumWet. Ser. C68 (1971) 122–127. from blood, IEEE Trans. Biomed. Engineering 39(5) (1992)

[9] H. Schmid-Schonbein, K.A. Kline, L. Heinich, E. Volger, T. 462–469.Fischer, Microrheology and light transmission of blood. III. [25] Y.F. Law, K.W. Johnston, H.F. Routh, R.S.C. Cobbold, OnThe velocity of red cell aggregate formation, Pflugers Arch. the design and evaluation of a steady flow model for Doppler354 (1975) 299–317. ultrasound studies, Ultrasound Med. Biol. 15(5) (1989) 505–

[10] M. Donner, M. Siadat, J.F. Stoltz, Erythrocyte aggregation: 516.Approach by light scattering determination, Biorheology 25 [26] S. Jones, D. Giddens, A simulation of trasit time effects in(1988) 365–375. doppler ultrasound signals, Ultrasound Med. Biol. 16(6)

[11] A. Gaspar-Rosas, G.B. Thurston, Erythrocyte aggregate (1990) 607–619.rheology by transmitted and reflected light, Biorheology 25(1988) 471–487.