European Journal of Ultrasound 8 (1998) 1–6

Clinical Paper

Variability of velocity and duration of microembolic signalsdetected by bigated transcranial Doppler sonography in carotid

endarterectomy

Martin Muller a,*, Xudong Pan b, Paul Walter c, Klaus Schimrigk c

a Department of Neurology, Uni6ersity Hospital of the Saarland, Oscar-Orth-Strasse 3, D-66421 Homburg/Saar, Germanyb Department of Surgery, Uni6ersity Hospital of the Saarland, Oscar-Orth-Strasse 3, D-66421 Homburg/Saar, Germany

c Department of Neurology, Affiliated Hospital of Qingdao Medical College, Jiangsu Road 16, 266003, Qingdao, PR China

Received 12 September 1997; received in revised form 29 January 1998; accepted 18 May 1998

Abstract

Objecti6e : To differentiate between gaseous and particular microemboli in carotid surgery one clinical approach isthe interpretation of the effective sample volume length (SVL). We investigated whether such a clinical interpretationis based on reproducible measurements. Methods : Microembolic signals (MES) recorded during carotid endarterec-tomy by a bigated transcranial Doppler device were analyzed off-line. In the two sample volumes, the duration andthe velocity of the MES were measured by two observers independently from each other twice within 2 weeks. TheSVL of the MES were calculated by multiplying duration with velocity. Results : In the anatomical proximal samplevolume 215 MES were recorded of which 203 (94.5%) were also present in the distal. The SVL medians of the MESwere 2.2–4.1 mm lower in the distal than in the proximal sample volume as a result of lower velocity and shorterduration of the MES in the distal sample volume. The median of the paired differences of the SVL was 0.2 mm(interquartile range: 0.0–1.2) in the proximal sample volume and 0.8 mm (0.2–1.8) in the distal sample volume forobserver 1, and 0.6 mm (0.4–2.2) and 0.9 mm (0.5–1.6) for observer 2. The median of the paired differences of theSVL between the observers was 1.4 mm (1.2–2.9) in the proximal sample volume and 1.6 mm (1.3–3.0) in the distal.Conclusion : The intra- and interobserver agreement on calculating SVL is good. However, the depth of insonationinfluences some features of embolic signals. © 1998 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Carotid endarterectomy; Microembolism; Stroke; Doppler sonography; Transcranial Doppler

* Corresponding author. Tel.: +49 6841 164113; fax: +49 6841 164137.

0929-8266/98/$ - see front matter © 1998 Elsevier Science Ireland Ltd. All rights reserved.

PII S0929-8266(98)00044-5

M. Muller et al. / European Journal of Ultrasound 8 (1998) 1–62

1. Introduction

Although many microembolic signals (MES)within the cerebrovascular circulation remainclinically silent, there is increasing evidence thatMES are a relevant marker for cerebral ischemicevents (Jansen et al., 1994; Smith et al., 1995;Muller et al., 1998). The ultrasound characteris-tics depend on the composition, size and proba-bly on velocity (Droste et al., 1994; Markus andHarrison, 1995; Smith et al., 1995). Using ultra-sound for an approach to differentiate betweenvarious embolic material, one has to keep inmind that a given ultrasound signal can be theresult of emboli which are composed of differentmaterial but are of different size (Markus andBrown, 1993). In carotid endarterectomy (CEA)particulate emboli and air-emboli may occur.Thus, small air-emboli may be indistinguishablefrom larger solid emboli. Because air-emboli aresuggested to rarely cause significant morbidity,and most neurological deterioration is believed tobe caused by particulate emboli (Gaunt et al.,1994; Smith et al., 1994), a differentiation be-tween these two embolic materials is, neverthe-less, desirable. Smith et al. (1995) based theinterpretation of embolic events on the effectivesample volume length (SVL) of MES. The SVL isthe product of the embolic duration in the samplevolume and its velocity. The SVL of air-emboliwas significantly larger than that of particulateemboli and allowed to classify 97% of the airemboli as such. In order to establish whether theclinical interpretation of such data is based onreliable measurements, we performed an intra-and interobserver study on the measurement ofvelocity, duration and SVL in MES occurringduring CEA.

2. Subjects and methods

The MES were collected from 16 patients (14male, two female, mean age 66 years, range 48–73 years), undergoing CEA due to severe stenosisof the internal carotid artery (more than 70%according to the European Carotid Surgery Trial;European Carotid Surgery Trialists’ Collabora-

tive Group, 1991). CEA was performed in a stan-dard manner with the use of normotensive,normo-cabnic general anesthesia, and TCD crite-ria for shunting (Jansen et al., 1994). The TCDmonitoring was performed with the use of a bi-gated TCD device (MultiDop X4, DWL, Sipplin-gen) from skin incision to wound closure andincluded blood velocity measurement and embolidetection. The probe was fixed with a metal headband so that the blood velocity wave form wasclearly present in both sample volumes. Eachsample volume had a length of 8 mm with aspatial distance of 5–6 mm between them. Thedepths of insonation ranged ranged between 58and 51 mm for the anatomical proximal samplevolume and between 52 and 46 mm for the distal.Sweep speed was set at 4 s, the used pulse repeti-tion frequency was 4.96 kHz. The threshold forsaving an embolic signal automatically by thesoftware was set on 4 dB above the backgrounddB value. The software saves on hard disc foroff-line analysis the blood velocity Doppler spec-trum, the dB value, the Raw Doppler amplitude/time display prior to the fast Fouriertransformation (FFT) processing, and the time ofoccurrence of each possible embolic event (Fig.1). The whole procedure was continuouslyrecorded on an audio/video tape.

MES were defined (Consensus Committee ofthe Ninth International cerebral HemodynamicSymposium, 1995) as signals unidirectional withinthe Doppler spectrum with a short duration (lessthan 0.3 s), and an amplitude of more than 3 dBabove the background signal. Each signal had tobe accompanied by a characteristic ‘chirping’sound. In addition to this definition, MES inCEA can be overloaded especially after clamprelease. Therefore, over loaded signals were in-cluded in our analysis. The dB value of an em-bouls is calculated by the TCD device in theproximal sample volume. The 4 dB threshold wasonly used to define a signal as embolic in nature.

The video tapes were reviewed, and only suchsignals were selected for off-line analysis onwhich both observers agreed that the signal was atrue MES. The time of the signal’s occurrencewas first determined on the video tapes, thereafterit was identified in the files on hard disc.

M. Muller et al. / European Journal of Ultrasound 8 (1998) 1–6 3

Fig

.1.

An

embo

lus

assa

ved

byth

eus

edT

CD

devi

ce.

Lef

tpa

rt:

next

toth

eco

lor-

code

dde

cibe

l(d

B)

scal

eth

eD

oppl

erve

loci

tysp

ectr

umaf

ter

fast

Fou

rier

tran

sfor

mat

ion

(FF

T)

ofea

chsa

mpl

evo

lum

eis

show

n(t

hepr

oxim

alsa

mpl

evo

lum

ew

ith

ade

pth

ofin

sona

tion

of54

mm

,th

edi

stal

wit

ha

dept

hof

48m

m).

The

embo

licsi

gnal

inte

nsit

yab

ove

the

back

grou

ndsi

gnal

(31

dB)

isca

lcul

ated

from

the

prox

imal

sam

ple

volu

me.

The

curs

orfo

rve

loci

tym

easu

rem

ent

(V)

isse

tat

the

uppe

rlim

itof

the

sign

alar

eaw

ith

the

high

est

inte

nsit

y(b

righ

test

colo

r-co

ded

area

).R

ight

part

:R

awD

oppl

eram

plit

ude/

tim

edi

spla

ypr

ior

toF

FT

proc

essi

ng.

The

curs

ors

are

set

atth

ebe

ginn

ing

and

the

end

ofth

esi

gnal

.T

hedu

rati

onof

the

sign

alis

43.7

ms

inth

epr

oxim

alsa

mpl

evo

lum

e,an

d38

.3m

sin

the

dist

al.

The

tim

ede

lay

the

embo

lus

need

edto

bere

cord

edin

the

dist

alsa

mpl

evo

lum

eis

26m

s.

M. Muller et al. / European Journal of Ultrasound 8 (1998) 1–64

Table 1Duration, velocity and sample volume length of the investigated microembolic signals

Duration (ms) Velocity (cm/s) SVL (mm)

1. Measurement28.0(4.8–110.0) 48.0(11.3–133.0) 12.1(1.8–62.4)Obs 1 Prox24.7(4.3–106.0) 42.0(11.3–118.0)Dist 9.9(1.8–57.0)Obs 1

ProxObs 2 26.8(5.0–108.0) 45.0(9.8–129.0) 11.1(1.9–57.0)22.6(2.7–93.6) 43.0(9.8–111.0) 8.2(0.9–51.6)DistObs 2

2. Measurement27.2(4.3–107.0) 47.1(12.1–133.0) 12.4(1.9–61.3)Obs 1 Prox23.5(3.7–96.6) 41.6(11.9–116.8)Dist 9.3(1.0–58.7)Obs 1

ProxObs 2 27.9(4.2–98.7) 46.5(9.8–129.0) 12.3(1.6–55.4)Obs 2 21.9(3.0–83.2)Dist 43.5(9.8–114.0) 8.2(1.3–64.0)

SVL, sample volume length; obs, observer 1 or 2; prox, proximal sample volume; dist, distal sample volume; ms, millisecond; mm,millimeter; cm/s, centimeter per second.Values are medians (minimum and maximum).

Within 2 weeks each observer measured twicethe velocity (in cm/s) and the duration of theMES (in milliseconds, ms) in both sample vol-umes as shown in Fig. 1. The intensity of theembolic signal is presented color coded within theDoppler spectrum. The color scale reaches fromblue to green, red, yellow and to white indicatingan increasing intensity in that order. For measur-ing velocity the cursor was set at the upper end ofthe area with the highest color coded intensitybecause in overloaded signals the highest intensitywithin the signal area can not be allocated asexactly as in signals with intensities remainingwithin the dynamic range of the spectral display(Fig. 1). The duration of each signal was mea-sured from its beginning to its end in the RawDoppler amplitude/time display as shown in Fig.1. From these measurements the SVL were calcu-lated by multiplying velocity with duration.

3. Data analysis

With respect to the consideration of Bland andAltman (1986), we analyzed the paired differencesof the measurements for estimating agreement.The results are reported as median and range oras median and interquartile range.

4. Results

A total of 215 MES were identified in theproximal sample volume of which 203 (94.5%)were also present in the distal. The median dBvalue of the MES was 23 dB (range: 5–47 dB).The median time delay of the MES between thetwo sample volumes was 10.20 ms (range: 0.70–47.80 ms, observer 1) and 9.00 ms (0.20–36.70ms, observer 2), respectively.

The MES showed a longer duration, a highervelocity, and, hence, a larger SVL in the proximalsample volume than that in the distal samplevolume at both times of measurement (Table 1).The median differences of the SVL between bothsample volumes ranged between 2.2 and 4.1 mm.

Considering the results of each observer, thepaired differences between the first and secondmeasurements were very small (Table 2) in bothsample volumes with the result that the SVL ineach sample volume was nearly equal at bothmeasurements (Table 1).

The interobserver paired differences (Table 2)for the duration were higher in the distal than inthe proximal sample volume, while the velocitymeasurements showed a reversed result. The dif-ference for the SVL was about 1 mm in bothsample volumes.

M. Muller et al. / European Journal of Ultrasound 8 (1998) 1–6 5

Table 2Paired differences for duration, velocity and sample volume length measured twice within 2 weeks for intraobserver agreement, andpaired differences between the first measurements of both observers for interobserver agreement

Paired difference

Duration (ms) Velocity (cm/s) SVL (mm)

Intraobserver agreementObs 1 0.30(−0.93–1.93)Prox 0.00(−0.90–0.50) 0.1(−0.5–1.1)

0.50(0.50–2.15) 0.00(−1.00–0.50)Dist 0.2(−0.3–1.0)Obs 1ProxObs 2 1.05(0.00–2.63) 0.00(−2.00–2.00) 0.5(−0.2–2.0)

Obs 2 1.25(−0.50–4.73)Dist 0.00(−1.90–2.90) 0.6(−0.2–1.9)

Interobserver agreement1.60(0.25–4.35)Prox 1.00(−2.00–4.00) 10(−0.8–2.8)2.10(0.00–8.00) 0.00(−2.00–2.00) 0.9(−0.1–3.8)Dist

SVL, sample volume length; obs, observer 1 or 2; dist, distal sample volume; prox, proximal sample volume; ms, millisecond; mm,millimeter; cm/s, centimeter per second.Values are median and interquartile range.

5. Discussion

Smith et al. (1995) used the Wigner distributionfor measuring the duration of a microembolicsignal, and were able to differentiate betweenparticulate and gaseous embolic material with anaccuracy believed to be useful for clinical interpre-tation of MES in CEA. In the TCD device weused the analysis of MES which is based on FFTanalysis. Compared with the Wigner analysis(Smith et al., 1994, 1995) the FFT analysis has apoorer time resolution and may therefore provide,some uncertainty for measuring the duration of amicroembolic signal within the Doppler spectrum.Although we achieved similar SVL results usingFFT analysis our SVL results cannot directly becompared with those of Smith et al. (1995) be-cause we measured the duration of a microem-bolic signal from its beginning to its end whileSmith et al. defined the duration of a microem-bolic signal as the time between a 10 dB thresholdat the onset and at the end of the signal. The aimof our study was the agreement on the measure-ments of the basic variables of a microembolicsignal and their effect on the resulting SVL. Wedid not intend to calculate from our results theSVL as a basis for clinical interpretation to differ-entiate between solid and gaseous embolicmaterial.

At both measurements of each observer theduration of the MES was longer and the velocityhigher in the proximal sample volume than in thedistal sample volume with the same result for theSVL. The SVL medians between the two samplevolumes differed between 2.2 and 4.1 mm (Table1). The very small paired differences between thetwo measurements of each observer favor thisresult. With respect to the relationship betweenvelocity and the duration of a microembolic signalour results confirm the finding of Smith et al.(1994), but are in contrast to the result of Drosteet al. (1994) who stated that the duration of amicroembolic signal decreases with an increasingvelocity. One possible explanation for this dis-crepancy could be the slight uncertainty in exactvelocity measurements in overloaded signalswhich we included in our study, while Droste etal. did not use such signals. A technical explana-tion can be that the sensitivity required to detectMES was higher in the proximal sample volumethan in the distal (Smith et al., 1994). Althoughwe took great care in the positioning of the twosample volumes one has to consider that the distalsample volume was not as accurately positionedin the blood stream as the proximal. In thiscondition MES with a lower signal intensity canshow a shorter duration than MES with a highsignal intensity (Smith et al., 1995). The use of

M. Muller et al. / European Journal of Ultrasound 8 (1998) 1–66

one ultrasound beam for originating two samplevolumes can lead to different angles of insonation,when the insonated vessel does not run straightahead as often found for the middle cerebralartery with the result that the velocities measuredin both sample volumes can differ. The TCDdevices providing only one sample volume facesthe same difficulties when the anatomical courseof the vessel is unfavorable for insonation. Whenusing SVL for differentiating solid from gaseousemboli our results indicate that the depth of in-sonation may be a relevant factor for classifyingemboli with respect to their composition.

The paired differences showed for all intra- andinterindividual measurements (Table 2) a centraltendency which indicates most differences aregathered between 0.1 and 1 mm with a slightlylarger difference in the intraobserver study com-pared to the interobserver study. Our results indi-cate a good intra- and interobserver agreementregarding the measurements of velocity and dura-tion and the calculation of a SVL. However, theydo not indicate whether the technique of Smith etal. (1995) for the differentiation between solid andgaseous emboli is useful enough for clinical prac-tise. Further studies are therefore necessary. Oneapproach could be the use of multiple gates in-stead of only two. Multiple gates placed along themiddle cerebral artery would allow to perform themeasurements in the gate in which the velocity ismaximum so that the angle of insonation is asclose to 0o as possible. Another approach couldbe the comparison of the technique of Smith et al.(1995) with the recently developed multi-fre-quency approach (Brucher and Russell, 1997) todifferentiate between solid and gaseous emboli.

References

Bland JM, Altman DG. Statistical methods for assessingagreement between two methods of clinical measurement.Lancet 1986i;307–310.

Brucher R, Russell D. Differentiation between gaseous andsolid microemboli using multi-frequency Doppler, Eur JUltrasound 1997;5 Suppl. 1:40–41 (Abstract).

Consensus Committee of the Ninth International CerebralHemodynamic Symposium: Basic identification criteria ofDoppler microembolic signals. Stroke 1995;26:1123.

Droste DW, Markus HS, Nassiri D, Brown MM. The effectof velocity on the appearance of embolic signals studiedin transcranial Doppler models. Stroke 1994;25:986–91.

European Carotid Surgery Trialists’ Collaborative Group:MRC European Carotid Surgery Trial: interim resultsfor symptomatic patients with severe (70–99%) or withmild (0–29%) carotid stenosis. Lancet 1991;337:1235–1243.

Gaunt ME, Martin PJ, Smith JL, et al. Clinical relevance ofintraoperative embolization detected by transcranialDoppler ultrasonography during carotid endarterectomy:a prospective study of 100 patients. Br J Surg1994;81:1435–9.

Jansen C, Ramos LMP, van Heesewijk JPM, Moll FL, vanGijn J, Ackerstaff RGA. Impact of microembolism andhemodynamic changes in the brain during carotid en-darterectomy. Stroke 1994;25:992–7.

Markus HS, Brown MM. Differentiation between differentpathological cerebral embolic material using transcranialDoppler in an in vitro model. Stroke 1993;24:1–5.

Markus HS, Harrison MJ. Microembolic signal detection us-ing ultrasound. Stroke 1995;26:1517–9.

Muller M, Behnke S, Walter P, Omlor G, Schimrigk K.Microembolic signals and intraoperative stroke in carotidendarterectomy. Acta Neurol Scand 1998;97:110–7.

Smith JL, Evans DH, Fan L, et al. Interpretation of em-bolic phenomena during carotid endarterectomy. Stroke1995;26:2281–4.

Smith JL, Evans DH, Fan L, Thrush AJ, Naylor AR. Pro-cessing Doppler ultrasound signals from blood borne em-boli. Ultrasound Med Biol 1994;20:455–62.

.