Application of Interhemispheric Index for Transcranial Doppler Sonography Velocity Measurements and Evaluation of Recording Time
Background and Purpose To validate the reliability of transcranial Doppler sonography velocity measurements in clinical settings, assessment of the reproducibility of repeated bilateral simultaneous measurements and the optimal recording time is needed. Our hypothesis was that interhemispheric indices would prove more valid than the absolute velocity measurements usually applied. The potential interference between ultrasound beams in bilateral samplings also needs evaluation.
Methods Thirty healthy volunteers were studied at rest within hours or with a 2-month interval between measurements. Absolute blood flow velocities and side-to-side indices between velocities obtained in the middle cerebral arteries were measured over a 30-second period by two independent examiners. The correlation coefficient (r) and the coefficients of variation of the difference between either absolute velocities (CV) or indices (CVi) were calculated. The beat-to-beat variation of the diastolic, systolic, and mean velocities was also recorded. For evaluation of ultrasound beam interference, measurements were performed with and without one probe unplugged.
Results In the interobserver study in which measurements were repeated within hours, r=.92, CV=8.8%, and CVi=4.1%. In the intraobserver study in which measurements were repeated with a 2-month interval, r=.8, CV=13.0%, and CVi=7.3%. A recording time of 30 seconds reduced CVi to 2.6%, whereas for absolute velocities 5-second recordings produced an acceptable variation. There was no significant interference between bilaterally placed probes.
Conclusions The introduction of interhemispheric indices improves interobserver and intraobserver reproducibility by approximately 50%. We recommend use of the index in clinical settings in which unilateral velocity changes are expected. For measurement of an interhemispheric index, a recording time of 30 seconds is recommended, whereas 5-second measurement periods yield a sufficient estimate of absolute velocities.
Transcranial Doppler sonography was introduced in 1982 for noninvasive and continuous monitoring of blood flow velocity in the basal cerebral arteries.1 Since then, the TCD technique has been widely used to evaluate regional cerebral blood flow changes in both physiological and pathological conditions, assuming that vessel diameter and perfusion territory remain unchanged with time. The technique has been applied to studies of cerebral autoregulation2 and CO2 reactivity3 and certainly adds another tool in the evaluation of patients with extracranial and intracranial stenosis, vasospasm after subarachnoid hemorrhage, acute stroke, arteriovenous malformations, head injury, brain death,4 and possibly other clinical conditions. In addition, the technique has proven valuable in clinical pharmacology trials testing the cerebral hemodynamic effects of drugs, eg, anesthetics,5 nonsteroidal anti-inflammatory drugs,6 and angiotensin-converting enzyme inhibitors.7
A major prerequisite for the use of the TCD technique in clinical settings is the reliability and reproducibility of measurements. Obviously, the assumption that cerebral blood flow velocity is constant during measurements and that all variations detected reflect the degree of insufficient reproducibility8 9 may not be valid, since physiological changes in blood flow velocity occur, ie, variations are caused by both insufficient reproducibility and physiological changes.
Therefore, the aim of this study was to evaluate the interobserver and intraobserver reproducibility of TCD velocimetry for repeated bilateral simultaneous measurements in the MCA under the assumption that side-to-side indices between the MCA blood flow velocities would be constant during and between measurements. Such indices, defined as the ratio between left and right or right and left absolute MCA velocities, were hypothesized to be relatively resistant against physiological systemic fluctuations in absolute blood flow velocity. In addition, spontaneous changes in MCA velocities with time were evaluated to determine the optimal recording time for both indices and absolute velocities. Beat-to-beat variations; the relationship between depth of the sample volume, intertemple distance, and probe position at the temporal window; and the relationship between probe position and absolute MCA velocities were also assessed. The potential interference between ultrasound beams occurring during bilateral recordings was also evaluated.
Subjects and Methods
Thirty volunteers (16 men and 14 women; median age, 45 years; range, 21 to 72 years) were examined. The subjects had no history of cerebrovascular disease and did not receive any medication. Our study was approved by the local ethics committee, and each individual provided written informed consent.
Blood flow velocity measurements were performed with 2-MHz pulsed-wave TCD equipment (Multi-Dop X hardware, DWL; version 7.40 of MDX TCD-7 software) able to make two-channel fast Fourier transformation calculations. The positioning of probes was maintained over each temple by an elastic headband.
Between September 1995 and February 1996, all subjects were examined twice (examinations 1 and 2). Each session lasted between 1.5 and 4 hours. Tea, coffee, alcohol, and smoking were not allowed on examination days. Before the first examination, the basic principles of the equipment were demonstrated, and each individual was helped to find his own MCA Doppler signal with a hand-held probe. We attempted to provide a calm environment with minimal acoustic distractions. Throughout the examinations, subjects were resting in the supine position; during measurements, subjects were requested to close their eyes, relax, and “think of nothing.” Subjects were not allowed to speak, nor were they spoken to. To maximize cardiac stability, subjects were resting at least half an hour before the first measurement was made.
We obtained Doppler signals from the MCA by placing each probe over the temple just superior to the zygomatic arch and adjusting its position and its angulation for a maximally reflected signal, characteristic of the MCA. The depth of the sample volume was increased from 50 mm by 2-mm steps until bidirectional flow appeared from the bifurcation of the internal carotid artery. Then the depth of the sample volume was decreased and the tilt of the probe was carefully adjusted to obtain a maximally reflected signal, characteristic of the M1 segment of the MCA. Signals were identified by means of the spectral screen display and an audible pitch. After we adjusted the axial length of the sample volume and gain to standardized values (10 and 8 mm, respectively), the spectral outline, ie, envelope, was ready for recording. Bilateral simultaneous waveform recordings were made over a period of at least 60 seconds. The depth of the sample volume and the probe position (at the anterior, middle, or posterior temporal window) were noted for each recording. The intertemple distance was measured by a slide gauge.
As shown in Table 1⇓, subjects underwent two series of measurements. In examination 1, subjects were investigated by two observers (A and B). Each of the observers placed a probe over a temple three times, and five bilateral simultaneous waveform recordings were made, ie, five indices were obtained. To avoid any bias, order of probe positioning was randomized so that the order of measurements was reversed in half of the cases.
An interval of 57 to 96 days passed between examinations 1 and 2. Subjects were reinvestigated by the same observers, A and B. Each observer placed a probe twice, and two recordings were done as shown in Table 1⇑. Each A4/A5 recording lasted at least 7 minutes and represented two separate measurement periods with 5 minutes in between and no change in probe position.
Subjects were asked not to help the observers during examinations. Once a probe had been fixed on one side, its wire was disconnected from the computer to ensure blinding and minimize ultrasound exposure while the second probe was positioned on the opposite side. The screen was covered during recordings, so that the exact relationship between velocities remained unknown to both observers.
For assessment of reproducibility and spontaneous evolution with time, the absolute MV and the side-to-side indices between MV of the left and the right MCA were considered.
To test for interference, the last A4/A5 measurement periods each included three intervals of 30 seconds, the last of which was performed in the left MCA (A4) only, ie, the wire of the right probe was disconnected.
Data and Statistical Evaluation
Evaluation of the hemispheric side-to-side difference in velocity was based on the first indices recorded (B1/A1). This approach seemed reasonable because the data showed that the observers on average measured equal velocities.
The data were used in four different studies, as discussed below. The potential interference between ultrasound beams occurring during bilateral recordings was also evaluated.
Interobserver Reproducibility Study
Evaluation of interobserver reproducibility was based on data obtained in examination 1. Duration of waveform recordings used in calculations was 30 seconds. Standard deviations and coefficients of variation were calculated with either indices or absolute velocities. SDi denotes the standard deviation of the differences within indices and was calculated between each pair of A1/A2 and B2/A2 and between each pair of A2/A1 and B1/A1 determinations. The coefficient of variation of the difference between indices was calculated as CVi=SDi/M*100%, where M is the mean of the indices measured by observer A. SD denotes the standard deviation of the differences within absolute velocities and was calculated between A1 and B2 and between A2 and B1 determinations. CV was calculated as CV=SD/M*100%, where M is the mean of the absolute velocity determinations of observer A. Linear regression analyses for comparison of absolute velocities (A1 versus B2 and A2 versus B1) were performed.
Intraobserver Reproducibility Study
Evaluation of intraobserver reproducibility was based on data obtained in examinations 1 and 2. To obtain an unbiased estimate of the intraobserver reproducibility, approximately 2 months elapsed between examinations. Duration of waveform recordings used in the calculations was 30 seconds. The intraobserver reproducibility was evaluated for both observers by means of the indices indicated in Table 1⇑ and was estimated as SDi and CVi. Furthermore, SD and CV for the measurements A1 versus A4 and B2 versus B4 were determined, and linear regression analyses for comparison of the absolute velocities were performed. SD, CV, and a linear regression analysis were applied to A1 and A3 determinations to assess short-term intraobserver reproducibility.
Beat-to-beat variation of MCA velocity parameters obtained in two successive cardiac cycles was determined in A3 recordings. Computerized calculations of MVs of the two A4/A5 measurement periods were done over 5, 10, 20, 30, 45, and 60 seconds, and corresponding velocities were compared. Evaluation was performed by calculating SDi, CVi, r, SD, and CV as previously described, and CVi was used to evaluate the short-term conservation of indices. Student’s t test was applied to compare the CVis and CVs obtained with different observation times.
Depth of Sample Volume Versus Head Size and Probe Position
A linear regression analysis of the correlation between the depth of sample volume and the head size determined as the intertemple distance was performed. For comparison of depth of sample volume, depth of sample volume divided by intertemple distance, and MV at the middle and posterior temporal windows, the Mann-Whitney test was applied.
A Mann-Whitney test was applied to test for differences within absolute velocities (A4) of the three 30-second intervals.
Two subjects had inappropriate temporal windows on either side and one on both sides. In total, 216 bilateral waveform recordings were obtained for data analysis. No artifacts with spike configuration appeared.
The averaged MVs were 74±15 cm/s on both the left and right MCA. The mean of left/right (B1/A1) indices was 1.00; the SD was ±16%. After exclusion of one subject who had a localized increase in MV and an index of 1.56, who was suspected of having a unilateral MCA stenosis, the SD was 11%.
Fig 1⇓ illustrates the correlation between absolute MCA velocities as measured by observers A and B on the left side.
Table 2⇓ summarizes the results of the interobserver reproducibility study. Correlations were statistically significant (P<.001). CVis of the left and the right MCA were of the same order of magnitude. As expected, CVi was comparable to SDi, since the mean index was close to 1. CVi was approximately half the value of CV.
Fig 2⇓ illustrates the correlation between absolute MCA velocities as measured by observer A on the left side, with approximately 2 months between measurements.
Table 3⇓ summarizes the results of the intraobserver reproducibility study based on either indices or absolute velocities obtained from the left MCA. The study yields an estimate of the CVi of roughly 7%. Again, CVi was approximately half the size of CV. Correlations were statistically significant (P<.001).
Statistical parameters of beat-to-beat variations are given in Table 4⇓. The best correlation, evaluated as the correlation coefficient and the CV, was found between MV values, followed by SV and DV, respectively. Correlation coefficients of SV, MV, and DV from heartbeat to heartbeat were all statistically significant (P<.001). Because of differences in mean values of SV and DV, the higher SD between SVs than between DVs is not reflected in the corresponding values of CV.
Variations of the differences between absolute velocities and indices of two measurement periods performed with an interval of 5 minutes, with no change in probe position, are illustrated in Figs 3⇓ and 4⇓. Differences between absolute velocities recorded over periods of 5 and 60 seconds did not differ significantly. Absolute differences between indices recorded over periods of 45 or 60 seconds were significantly smaller than differences recorded over a period of 5 seconds (Student’s t test; P<.05). Absolute differences between indices recorded over a period of 30 seconds were not significantly smaller than differences recorded over a period of 5 seconds (Student’s t test; P=.06). When recordings were done over a period of 30 seconds, the variation of the differences, evaluated as CVi, reached a level within acceptable limits (2.6%). This value was not statistically different from the CVi resulting from 60 seconds of recording time. The corresponding SDi was 0.03, and the corresponding variation of the differences between absolute velocities on the left side yielded values for r, SD, and CV of .91, 5.8 cm/s, and 8.5%, respectively.
There was a statistically significant correlation between depth of sample volume on the left side and intertemple distance (n=16, r=.72, P<.005) according to the equation Depth of Sample Volume=0.47 ×Intertemple Distance (mm)−13.8 mm
The number of probe positions at the anterior, middle, and posterior temporal windows was 1, 14, and 12, respectively. Insonation through the middle temporal window was performed with a mean depth (52 mm) that was lower than that through the posterior window (55 mm), although the difference was not significant; nor was the difference between ratios of mean depths and intertemple distance significant. Mean velocities showed a trend toward higher values when probes were fixed at the middle (average MV, 76 cm/s) compared with the posterior (average MV, 69 cm/s) temporal window, although the difference was not statistically significant.
The absolute velocity on the left MCA was not significantly different in the presence compared with the absence of contralateral insonation.
In this study, MVs in the left and right MCA were both 74±15 cm/s. This level is somewhat higher than those observed by Aaslid et al (62±12 cm/s)1 but corresponds well to those obtained by Sorteberg et al (73±11 cm/s).10 Our SD of the average side differences in flow velocity was 16%. After exclusion of one subject suspected to have a unilateral MCA stenosis, the SD was 11%, which was in better agreement with the SD of Sorteberg et al (7%).
The CV obtained in two measurement periods with 5 minutes between measurements and no change in probe position is an estimate of the spontaneous variation of MV with time. Based on 30-second recordings, this variation was as high as 8.5%. This value is similar to the interobserver reproducibility estimated as the CV, which was 8.8%. Thus, when the probe is carefully positioned, interobserver variation is almost entirely caused by spontaneous fluctuations with time, with only a negligible contribution from interobserver disagreement.
The interobserver reproducibility for side-to-indices, CVi, was 4.1%, ie, approximately half the value of the corresponding CV. Similarly, in the intraobserver reproducibility study, CVi and CV were 7.3% and 13.0%, respectively. The smaller variation of MCA indices than of absolute velocities is probably caused by physiological fluctuations in absolute blood flow velocities being reflected bilaterally, eg, systemic changes in arterial CO2 partial pressure3 or changes in cardiac output and mean arterial blood pressure. Whenever measurements of indices are relevant, the need for continuous end-tidal CO2 partial pressure and mean arterial blood pressure monitoring during measurements is minimized, which is beneficial because monitoring of end-tidal CO2 partial pressure may have an unpredictable influence on blood flow velocity and correction for changes in arterial CO2 partial pressure may be insecure11 or insignificant for recordings at rest.10 Furthermore, side-to-side indices are resistant to changes in hematocrit (known to be inversely related to the MCA velocity parameters12 ) between examinations, and possible diurnal variations in blood flow velocity were proposed by Demolis et al.13 On the other hand, indices may be influenced by hemispheric blood flow velocity shifts induced by cognitive tasks.14 15 Since side-to-side indices between absolute velocities are resistant to physiological changes reflected bilaterally, the CVis are far better reproducibility estimates than CVs. We therefore recommend the performance of simultaneous bilateral measurements and the use of side-to-side indices whenever unilateral velocity changes are expected. The use of indices would be of potential value in the evaluation of patients with conditions such as extracranial and intracranial stenosis, vasospasm after subarachnoid hemorrhage, and acute stroke.
Table 5⇓ summarizes the reproducibility of MCA absolute velocity reported in published TCD studies. Our interobserver reproducibility parameters are in accordance with those of Demolis et al13 and Maeda et al.8
An unbiased estimate of intraobserver reproducibility can only be obtained if the examiner is unaware of the result obtained from the previous examination. Therefore, a period of approximately 2 months between examinations was allowed to pass. It should be emphasized that the intraobserver reproducibility variance includes contributions caused not only by the examiner but also by the time interval between examinations and the duration of recordings. When the time interval between examinations was approximately 2 hours, our intraobserver reproducibility was 6.9%, whereas it increased to 13% when this time interval was approximately 2 months.
Since it is unlikely that the intraobserver disagreement (CV of observer A=13.0%) exceeds the interobserver disagreement (CV of the left MCA=8.8%), changes in MV during the 2-month period must account for the difference. Long-term dependent changes may explain why our reproducibility corresponds well to the findings of Maeda et al8 but less well to those of others9 13 16 (Table 5⇑). Alternatively, both observers may have experienced a change in their skills between examinations, but this is less likely, since the average MV did not differ significantly between examinations 1 and 2. Since measurements were performed at different times during examination days, diurnal variations might contribute to intraobserver disagreement only.
For absolute velocities obtained in two measurement periods with 5 minutes between measurements at unchanged probe positions, the CV did not change significantly whether the recording time was 5 or 60 seconds (Fig 3⇑). Therefore, after 5 seconds a sufficiently reliable measure of the absolute velocity seemed to be reached and prolongation of the measurement period did not lead to an improved estimate. With a recording time of 30 seconds, our findings (r=.91; SD=5.8 cm/s; CV=8.5%) were in agreement with the findings of Demolis et al13 (r=.90; SD=4.8 cm/s). In contrast, the CVi showed a decline with increasing observation time (Fig 4⇑). A recording time of 30 seconds seems, however, to yield a satisfactory CVi, which was not statistically different from the CVi resulting from a recording time of 60 seconds. Still, all CVis were lower than the CVs independent of recording time.
The beat-to-beat variation was smallest for absolute MVs, followed by SVs and DVs, respectively. To our knowledge, this is the first report of beat-to-beat variation on TCD velocimetry.
A significant correlation between depth of sample volume and intertemple distance was found and, as noted by Vriens et al,11 when depth of insonation is evaluated, the head size must be taken into consideration. The maximum reflected signal was found when insonation was performed through the middle temporal window (52%) or through the posterior temporal window (44%). This emphasizes the need to scan these windows systematically during examinations.
In conclusion, the introduction of MCA interhemispheric indices improves interobserver and intraobserver reproducibility by approximately 50%. We recommend the performance of simultaneous bilateral measurements and the use of the index in clinical settings in which unilateral velocity changes are expected. For measurement of the index, a recording time of approximately 30 seconds is recommended, whereas 5-second measurement periods yield a sufficient estimate of absolute MCA blood flow velocities. A significant interference between ultrasound beams could not be demonstrated.
Selected Abbreviations and Acronyms
|CV||=||coefficient of variation of differences between absolute velocities|
|CVi||=||coefficient of variation of differences between indices|
|DV||=||diastolic flow velocity|
|MCA||=||middle cerebral artery|
|MV||=||mean flow velocity|
|SD||=||standard deviation based on absolute mean velocities|
|SDi||=||standard deviation based on indices|
|SV||=||systolic flow velocity|
|TCD||=||transcranial Doppler sonography|
This study was supported by the Danish Health Research Council.
- Received September 12, 1996.
- Revision received January 10, 1996.
- Accepted January 13, 1997.
- Copyright © 1997 by American Heart Association
Larsen FS, Olsen KS, Hansen BA, Paulson OB, Knudsen GM. Transcranial Doppler is valid for determination of the lower limit of cerebral blood flow autoregulation. Stroke. 1994;25:1985-1988.
Thiel A, Zickmann B, Zimmermann R, Hempelmann G. Transcranial Doppler sonography: effects of halothane, enflurane and isoflurane on blood flow velocity in the middle cerebral artery. Br J Anaesth. 1991;68:388-393.
Markus HS, Vallance P, Brown MM. Differential effect of three cyclooxygenase inhibitors on human cerebral blood flow velocity and carbon dioxide reactivity. Stroke. 1994;25:1760-1764.
Sorteberg W, Langmoen IA, Lindegaard K-F, Nornes H. Side-to-side differences and day-to-day variations of transcranial Doppler parameters in normal subjects. J Ultrasound Med. 1990;9:403-409.
Vriens EM, Kraaier V, Musbach M, Wieneke GH, van Huffelen AC. Transcranial pulsed Doppler measurements of blood velocity in the middle cerebral artery: reference values at rest and during hyperventilation in healthy volunteers in relation to age and sex. Ultrasound Med Biol. 1989;15:1-8.
Brass LM, Pavlakis SG, DeVivo D, Piomelli S, Mohr JP. Transcranial Doppler measurements of the middle cerebral artery: effect of hematocrit. Stroke. 1988;19:1466-1469.
Demolis P, Chalon S, Giudicelli J-F. Repeatability of transcranial Doppler measurements of arterial blood flow velocities in healthy subjects. Cli Sci. 1993;84:599-604.
Droste DW, Harders AG, Rastogi E. A transcranial Doppler study of blood flow velocity in the middle cerebral arteries performed at rest and during mental activities. Stroke. 1989;20:1005-1011.
Rihs F, Gutbrod K, Gutbrod B, Steiger H-J, Sturzenegger M, Mattle HP. Determination of cognitive hemispheric dominance by ‘stereo’ transcranial Doppler sonography. Stroke. 1995;26:70-73.