Share this article on:

Forearm blood flow by Doppler ultrasound during rest and exercise: tests of day-to-day repeatability


Medicine & Science in Sports & Exercise: September 1996 - Volume 28 - Issue 9 - p 1144-1149
Basic Sciences/Regulatory Physiology: Original Investigations

The between-day repeatability of simultaneous measures of brachial artery diameter (D) (echo Doppler) and mean blood velocity (MBV) (pulsed Doppler) was tested during rest and exercise. On 3 separate days, six volunteers performed one trial of 1-min rest followed by a step increase in dynamic handgrip exercise for 4 min which required the lifting and lowering of a 4.4-kg weight(≈8-12% MVC) in a 1s/2s (work/rest) cadence. Measures for MBV and D were collected continuously on a beat-by-beat basis during the transition from rest to end exercise. The mean rest values over one min, and single data points at 30, 60, 120, and 240 s of exercise were extracted from the time series data. At all exercise time points, MBV was greater than rest (P < 0.05), but these levels were not different across test days. Arterial D at all exercise time points ranged from 3.8 ± 0.1 mm to 4.1 ± 0.1 mm(mean ± SEM) and did not differ from rest (3.9 ± 0.1 mm)(P > 0.05), nor did D differ between days. The mean between-day coefficient of variation for D was 4.08 ± 0.7% at rest and ranged from 2.90 ± 0.4% to 3.96 ± 0.5% during exercise. The coefficient of variation for MBV was 13.2 ± 2.6% at rest and reached 20.2 ± 3.1% during the final min of exercise; the exercise variability was reduced to 14.9 ± 2.4% by averaging MBV over 3 s (the duration of a contraction/relaxation duty cycle) (P < 0.05) with no further advantage of averaging over ten 60-s sample periods. The data indicate that, for the six subjects tested, Doppler ultrasound measures of arterial MBV and diameter during both rest and exercise were reproducible across different test days and can be used as a reliable, noninvasive means of testing hypotheses pertaining to blood flow control.

Department of Kinesiology, University of Waterloo, Waterloo, ON Canada N2L 3G1

Submitted for publication January 1996.

Accepted for publication June 1996.

This work was supported by the Natural Sciences and Engineering Research Council of Canada. Kevin Shoemaker was the recipient of an Natural Sciences and Engineering Research Council of Canada Graduate Scholarship.

Address for correspondence: R. L. Hughson, Ph.D., Dept. of Kinesiology, University of Waterloo, Waterloo, ON Canada N2L 3G1;

Understanding the mechanisms responsible for the hyperemia at the onset of human muscular exercise depends on the ability to obtain reliable quantitative measurements during the exercise. The standard for measurement of forearm blood flow has been strain gauge plethysmography, but this requires pauses in the muscle contraction pattern, and it also requires that the limb be elevated above the heart to facilitate venous drainage. Doppler ultrasound provides a noninvasive method of blood velocity quantification with beat-to-beat resolution of blood velocity. Therefore, Doppler ultrasound has recently become an attractive modality for studies of blood flow dynamics with exercise (2,8,11,14,17). When this method is combined with knowledge of vessel cross-sectional area, the quantitative blood flow can be evaluated.

The major limitations to date with Doppler techniques are that diameter measurements have been obtained only at rest(2,8,12,17) and that indications of the reproductibility of the technique during rest-to-exercise transitions have not been presented. Because there is substantial beat-to-beat variability in blood velocity (1,2,9,13) it is important to know how this affects estimates of blood flow and the reproducibility of the method when applied to studies that investigate the mechanisms of exercise-induced hyperemia. In this study we report on measurements of continuous pulsed Doppler ultrasound to determine mean blood velocity (MBV), and on continuous echo Doppler imaging of the brachial artery, to evaluate cross-sectional area during rest and moderate levels of dynamic handgrip exercise on three separate test days.

Back to Top | Article Outline


Subjects. Four male and two female volunteers participated in this study. The subjects were 24 ± 1.2 yr of age, 1.81 ± 0.03 m in height, and weighed 74.0 ± 4.3 kg (mean ± SE). The subjects were free of any cardiovascular disease as determined by a medical screening form. After receiving a complete description of the experimental protocol and potential risks, each volunteer provided consent to the testing procedures that had been approved by the Office of Human Research at the University of Waterloo.

Experimental design. Each subject performed dynamic handgrip exercise that required the lifting and lowering of a 4.4-kg weight over a distance of 5 cm; this weight was 8-10% of each male's, and 12-13% of each female's, maximal isometric voluntary handgrip contraction. The exercise was performed in the supine position to diminish the contribution of changes in cardiac output and arterial pressure to the measured blood flow response. The exercising arm was elevated 50° above the heart and was stabilized on a supporting platform. An exercise cadence of 1 s/2 s (work/rest) was maintained with approximately 0.5 s of each contraction performing for the concentric phase and 0.5 s doing the eccentric work. With this workload and cadence the resulting work rate was 1.44 W.

Prior to the study, each subject was familiarized with the testing procedures and the exercise protocol by performing two practice sessions. During these practice sessions the experimenter also determined the optimal positions for the imaging and Doppler probes; these locations were marked for future reference. The experimenter then practiced with the subject during both orientation sessions to learn the pattern of arterial movement, if any, during the exercise. The importance of arm stability at all times, and muscle relaxation between contractions, was stressed.

The subjects reported to the laboratory at the same time of day on each of three separate days. A single exercise trial was performed on each day following ≈20 min of supine rest. Each trial included 1 min of rest followed by a step change in work rate which lasted for 4 min. To diminish the anticipatory responses to exercise, the subjects were not aware of the time in any trial. All measurements were made by the same experienced operator. Room temperature was 21-22°C. Subjects had not eaten for 2-3 h and had abstained from both alcohol and caffeine ingestion for 24 h prior to any trial.

Data collection and analysis. Heart rate (Cambridge Model VS4 electrocardiograph), MBV (pulsed Doppler velocimetry), and arterial diameters(Echo Doppler) were collected continuously during each exercise trial. MBV was determined from the spectra of a single-gated pulsed-Doppler ultrasound signal(Model 500V, Multigon Industries, Mt. Vernon, NY). A flat 4.0-MHz probe was fixed to the skin over the brachial artery a distance of ≈9 cm proximal to the medial epicondyle. This positioning of the Doppler probe was necessary to prevent acoustic interference with the imaging probe which was placed ≈6 cm distally. Brachial artery diameters were not different between the two measurement sites. The angle of the 4.0-MHz transducer relative to the skin was 45° and the depth and gate of the ultrasound band were adjusted to insonate the total width of the artery.

The analysis of the Doppler shift spectra has been described in detail previously (11). Briefly, the Doppler shift signal was analyzed by a quadrature audio demodulator, which provided instantaneous MBV in real time. The signal was calibrated by an electronically generated phase shift of 1 mV that allowed conversion of the demodulated data to MBV values calculated and stored on a computer-based system at 100 Hz.

Blood velocity is calculated as Equation

where V is velocity, fD is the Doppler shift (MHz) (i.e., the difference between the frequencies of the transmitted (ft) and reflected ultrasound), C is the velocity with which ultrasound travels in biological tissues, and θ is the angle of insonation.

The raw MBV data were subjected to a moving average procedure over three cardiac cycles. Beat-by-beat MBV values were then calculated as the average of the instantaneous MBV profile over each cardiac cycle. For each trial, the resting MBV value was determined as the mean of all values collected during the initial 1-min segment. Exercise MBV were derived from the beat-by-beat data at the 30, 60, 120, and 240 s time points.

The brachial artery was imaged continuously during each trial by a Diagnostic Ultrasound System (Toshiba Model SSH-1400A) operating in B-mode. A 7.5-MHz linear probe was positioned over the vessel in the region of the antecubital fossa using visual feedback to verify the optimal placement. The real-time image was saved on videotape for analysis.

Arterial diameters were always made from an image frozen at end diastole and always during the rest phase between muscle contractions. This was facilitated by an ECG trigger that updated the recorded arterial image at a point corresponding with the QRS complex. The measurement markers were set to move in increments of 0.1 mm. The mean resting diameter was the average of measurements made at 20, 35, and 50 s of the initial rest period. Exercise diameters were obtained at 30, 60, 120, and 240 s following the onset of contractions to correspond with beat-by-beat MBV measures. Each diameter value was the mean of three measurements. Resting forearm blood flow[MBV·(πr2)] was calculated from the average MBV and diameter values derived from the minute of rest preceding the exercise. Exercise blood flow was calculated from the individual MBV and diameter estimations at the respective time points.

The exercise MBV data were analyzed further to investigate the physiological effects of muscle contraction and relaxation on the MBV variability. We reasoned that the magnitude of contraction-induced variation should be reduced by averaging over time. Also, we wanted to determine whether including one contraction and relaxation cycle (i.e., a 3-s time bin) could optimally reduce MBV beat-by-beat variability. Therefore, during the period between 3 and 4 min of exercise, the MBV values were averaged over 3, 5, 10, 30, and 60 s windows with each segment beginning at the 3-min time point.

Pulsed Doppler calibration. In vitro calibration of the pulsed Doppler unit was performed by measuring the velocity of porcine blood, heated to 40°C, as it flowed through Tygon plastic tubing at a constant rate. Tubes of two diameters (1.8 and 4 mm) were used to produce a range of velocities. A syringe pump produced known flow rates which ranged from 20 ml·min-1 to 140 ml·min-1. From the nominal tube diameter and flow rates the actual blood velocity could be calculated. Repeated measures were collected at each of seven flow rates. The pulsed Doppler measures were then correlated with the known velocities to determine a calibration equation.

Statistics. The between-day variability for MBV, arterial diameter, and calculated blood flow was determined in two ways: First, the absolute values at the chosen time points were compared for statistically significant differences. Second, a coefficient of variation[(SD/mean)·100) was calculated for each subject at each of the selected times across the three days. The effect of exercise and testing day on MBV, arterial diameter, and blood flow values, were determined separately by a one-way analysis of variance (ANOVA) for repeated measures (Statistical Analysis System). The effect of time-averaging the exercise MBV data was also analyzed by one-way ANOVA with time-bin size as the dependent variable. The effect of averaging the MBV data over various durations was determined by one-way ANOVA. The level of significance was set at P < 0.05 and any differences were further analyzed with Student-Newman Keulspost-hoc test. Data are presented as mean ± SE.

Back to Top | Article Outline


Mean blood velocity. Compared with rest, MBV was increased with exercise (P < 0.05), but these levels were not different across days (P > 0.05) (Fig. 1 and Table 1). At rest the mean coefficient of variation for MBV was 13.1 ± 2.6%(Table 1). During exercise the mean coefficient of variation ranged from 13.2% at 120 s to 20.2% at 240 s of exercise. Compared with the single beat MBV value at 240 s, time-averaging the MBV between 3-4 min of exercise over 3 s (i.e., the duration of the contraction/relaxation cycle), reduced the exercise variability to 14.9 ± 2.4% (P< 0.05). With longer periods of time-averaging (5,10,30, and 60 s) exercise, MBV variability ranged from 9.77 to 11.6% (Table 2). There was no difference between these time averaging methods(P > 0.05).

Brachial artery diameter. Brachial artery diameters were 3.9± 0.1 mm at rest and were not different at 30, 60, 120, and 240 s of exercise (P > 0.05; Table 3), nor were they different between test days (P > 0.05). The mean coefficient of variation for D was 4.1 ± 0.7% at rest and ranged from 2.9 to 4.0% during exercise (Table 3).

Blood flow. Calculated blood flows were greater during exercise compared with rest (P < 0.05) but were not different between test days at any measured time point (P > 0.05)(Table 4). The coefficient of variation for blood flow between days was 16.7 ± 3.8% at rest and ranged from 14.8 to 21.0% during exercise. There were no differences in this indicator between sample times (P > 0.05).

Pulsed Doppler calibration. In vitro correlation of the pulsed Doppler estimates of the known blood velocity indicated a highly linear relationship (r = 0.99) with a slope of 0.9 and a y intercept of 2.63 cm·s-1.

Back to Top | Article Outline


Major limitations of ultrasound methods for blood flow determination during dynamic exercise have been the inability to obtain accurate and reproducible measures of arterial diameter and the highly variable blood velocity profile. As a consequence, previous reports have assumed that resting diameters are representative of exercise values (2,8,17) or that diameters have been calculated from the relationship between forearm blood flow, as measured by strain gauge plethysmography and arterial MBV as measured by continuous-wave Doppler (12). In the current study we provide evidence that measures of both brachial artery MBV and diameter by ultrasound can be obtained simultaneously during rest and dynamic handgrip exercise. Furthermore, arterial diameter estimates by direct sonography were highly reproducible during both rest and exercise with within-subject variability ranging from 2 to 4%. The MBV responses at rest and exercise were not different between days but were subject to greater levels of within-subject variability. Consequently, blood flow calculated from the product of MBV and cross-sectional area was not different between days.

Previously, trial-to-trial variability of femoral artery blood flow was reported to be ≈17% when measured by ultrasound under conditions of supine rest (13). These previous values compare closely with the day-to-day reproducibility of brachial artery flow observed under conditions of both rest and exercise in the current study. On the basis of a close correlation between the pulsed Doppler MBV and the known blood velocity through the Tygon tubes and previous reports of close relationships between blood flow measures by Doppler ultrasound and other techniques(1,5,7,10,15,16,18), it was assumed that the pulsed Doppler velocity estimate accurately represented the arterial MBV observed during the exercise. Therefore, the main sources of the between-day variation in blood flow observed here and by others(1,13) may involve errors in the estimation of arterial diameter, errors in the estimation of the insonation angle for MBV measures (1,4), real variation due to muscle contraction (2), or moment-by-moment dynamic control of arterial perfusion pressure (6) and downstream arteriolar tone (9).

Methodological sources of variability. Accurate diameter measures are of paramount importance because flow is a function of the squared vessel radius; any variations in diameter estimation will be magnified in the flow calculations. Since diameter measures were made with a hand-held probe in this study, operator error must be considered as a factor influencing accurate determination of vessel diameter. With the arm support apparatus and exercise modality used in this study, there was minimal arm movement; however, small but predictable movements of the brachial artery may occur with forearm contractions. Therefore, the imaging probe must be free to move with the artery to maintain an optimal image. This was best afforded by having the probe held by an experienced operator rather than a rigid mechanical device. Further attempts to diminish errors in diameter determination were made by practicing with each subject prior to beginning the study and by taking the average of three measures at each time point. A key indication that our imaging and data collection were consistently optimal for this application was the low and relatively constant level of variation between days.

A further source of diameter variation may be operator error in the placement of the measurement calipers. The measurement sensitivity of the Toshiba imaging system was set at 0.1 mm. For diameters in the range of 4.0 mm, a 0.1-mm error in placement of the marker could result in ≈2.5 - 3% of the variation. Therefore, measurement error could account for the measured variation in the brachial artery diameters.

To attenuate the effect of errors in the angle of insonation for velocity measures, the flat-pulsed Doppler probe was fixed to the skin over the brachial artery as it passes along the biceps muscle. Therefore, the angle of insonation was believed to be constant at 45°. Echo Doppler imaging confirmed that the brachial artery runs parallel to the skin's surface at this location and that during exercise contraction-induced fluctuations in the insonation angle were ≤5°. Therefore, the maximal error in MBV calculation, due to contraction-induced errors in estimation of the insonation angle, was 12% (see Fig. 2).

Finally, improper alignment of the ultrasound beam with the artery and Doppler signal processing and frequency estimation (4) may contribute to variability in the estimation of MBV. Frequency estimation can be affected by random noise associated with summing a very large number of independently scattered signals to which is added a heterogenous range of transmitted ultrasound frequencies. This random noise was reduced by coherent averaging of two or three repeated trials. Also, interference from vessel walls and surrounding tissues may augment the broadband nature of the Doppler shifted signal. High-pass filtering was used to attenuate any slow velocity interference of wall movement to the Doppler shift signal. These problems were further minimized by the use of both audio and visual feedback of the Doppler shift spectra to maintain signal strength.

Physiologic sources of variability. Eriksen et al.(2) observed substantial differences in calculated blood flow from one trial of mild knee extension exercise to the next and indicated that such variability may not rest with the methods of measurement but reflect true physiological differences in blood flow. During contraction of skeletal muscle, blood flow is temporarily stopped and is often reversed(17). Subsequent relaxation allows a rapid and large surge of blood through the artery, the magnitude of which is partially dependent on the phase of the cardiac cycle (near systole or diastole) at the time of relaxation.

If the variability in MBV during exercise was attributable to the effect of muscle contraction and relaxation, then averaging the MBV data over longer durations of time should reduce the variance. Averaging the end exercise MBV data over the duration of the contraction/relaxation cycle (3 s) reduced the day-to-day variation of the beat-by-beat measures from ≈20% to ≈15%, closely matching the variation observed at rest. Averaging the MBV data over 5-60 s minimally improved the day-to-day variability compared with the 3-s time bins.

The variable flow observed at rest (9,13) and exercise (2) might be a consequence of modulations in vascular smooth muscle tone (3). Such random and heterogeneous variations in microvascular tone may govern the beat-by-beat MBV variability observed in human conduit arteries while at rest(1,2,9,13).

Implications and recommendations. Stacey-Clear et al.(13) noted that the trial-to-trial blood flow variability of 17% was reduced to 12% when the first trial was excluded from the data analysis. Their (13) data were collected under resting conditions and they argued that well-practiced and rested subjects were of paramount importance to accurate blood flow measures by ultrasound. Based on our experiences with Doppler measures of blood flow to date, we have found that the main requirements for accurate and reproducible flow detection at rest are a relaxed and well-practiced subject, an experienced Doppler operator, and ambient temperatures that are cool enough to attenuate the periodic opening and closing of skin anastomoses (9). Additional requirements for ultrasound measures of blood flow during exercise include limb stability and optimal relaxation of the working limb between contractions. Also, the operator must know the pattern of arterial movement, if any, that occurs with muscular contraction.

Errors in determining arterial diameter and insonation angle are considered to be random in nature (4). Therefore, the magnitude of within-subject variability between days should be reduced by convergent averaging of two or more repeated trials.

Summary. Although the tests were performed on a limited number of subjects, the results of the present study support the use of Doppler ultrasound methods for testing hypotheses related to blood flow control during exercise. These data also support the use of Doppler techniques for the assessment of both the progression of cardiovascular disease and for the effectiveness of interventions as they pertain to that disease. A single diameter measure at the required time point during exercise was as reliable as the average of three measures taken at rest. Averaging beat-by-beat MBV data over the duration of a contraction/relaxation cycle reduced the exercise-induced variability optimally while maintaining sufficient time resolution for time course studies. In conclusion, measures of conduit artery MBV and diameter by pulsed and echo Doppler, respectively, can be collected simultaneously during both rest and dynamic exercise and can be reproduced reliably from one day to the next.

Figure 1- The mean blood velocity (MBV) responses to dynamic handgrip exercise were not different across 3 different test days. Data tracings are the mean of six subjects performing one trial each with data points averaged (A) over 1 s time bins or (B) over 3-s time bins to include both phases of the contraction/relaxation duty cycle.

Figure 1- The mean blood velocity (MBV) responses to dynamic handgrip exercise were not different across 3 different test days. Data tracings are the mean of six subjects performing one trial each with data points averaged (A) over 1 s time bins or (B) over 3-s time bins to include both phases of the contraction/relaxation duty cycle.

Figure 2- The calculated error in mean blood velocity (MBV) is shown for a given error in insonation angle if 45° is assumed. Over-estimation of the angle produces errors of greater magnitude than does underestimation.

Figure 2- The calculated error in mean blood velocity (MBV) is shown for a given error in insonation angle if 45° is assumed. Over-estimation of the angle produces errors of greater magnitude than does underestimation.

Back to Top | Article Outline


1. Chauveau, M., B. Levy, J. F. Dessanges, E. Savin, O. Bailliart, and J. P. Martineaud. Quantitative Doppler blood flow measurement method and in vivo calibration. Cardiovasc. Res. 19:700-706, 1985.
2. Eriksen, M., B. A. Waaler, L. Walloe, and J. Wesche. Dynamics and dimensions of cardiac output changes in humans at the onset and at the end of moderate rhythmic exercise. J. Physiol. 426:423-437, 1990.
3. Faber, J. E., P. D. Harris, and I. G. Joshua. Microvascular response to blockade of prostaglandin synthesis in rat skeletal muscle. Am. J. Physiol. 243:H51-H60, 1982.
4. Gill, R. W. Measurement of blood flow by ultrasound: Accuracy and sources of error. Ultrasound Med. Biol. 11:625-641, 1985.
5. Guldvog, I., M. Kjaernes, M. Thoresen, and L. Walloe. Blood flow in arteries determined transcutaneously by an ultrasonic Doppler velocity meter as compared to electromagnetic measurements on the exposed vessels. Acta Physiol. Scand. 109:211-216, 1980.
6. Hughson, R. L., A. Maillet, G. Dureau, Y. Yamamoto, and C. Gharib. Spectral analysis of blood pressure variability in heart transplant patients. Hypertension 25(part 1):643-650, 1995.
7. Levy, B. I., W. R. Valladares, A. Ghaem, and J. P. Partineaud. Comparison of plethysmographic methods with pulsed Doppler blood flowmetry. Am. J. Physiol. 236:H899-H903, 1979.
8. Leyk, D., D. Essfeld, K. Baum, and J. Stegemann. Early leg blood flow adjustment during dynamic foot plantarflexions in upright and supine body position. Int. J. Sports Med. 15:447-452, 1994.
9. Lossius, K., M. Eriksen, and L. Walloe. Fluctuations in blood flow to acral skin in humans: Connection with heart rate and blood pressure variability. J. Physiol. 460:641-655, 1993.
10. Lubbers, J., P. J. L. M. Bernink, G. J. Barendsen, and J. W. Van Den Berg. A continuous wave Doppler velocimeter for monitoring blood flow in the popliteal artery, compared with venous occlusion plethysmography of the calf. Pfluegers Arch. 382:241-248, 1979.
11. Shoemaker, J. K., L. Hughson. and R. L. Hughson. Cardiorespiratory kinetics and femoral artery blood velocity during dynamic knee extension exercise. J. Appl. Physiol. 77:2625-2632, 1994.
12. Sinoway, L. I., C. Hendrickson, W. R. Davidson, S. Prophet, and R. Zelis. Characteristics of flow-mediated brachial artery vasodilation in human subjects. Circ. Res. 64:32-42, 1989.
13. Stacey-Clear, A. and P. J. Fish. Repeatability of blood flow measurement using multichannel pulsed Doppler ultrasound. Br. J. Radiol. 57:419-420, 1984.
14. Toska, K. and M. Eriksen. Peripheral vasoconstriction shortly after onset of moderate exercise in humans. J. Appl. Physiol. 77:1519-1525, 1994.
15. Tschakovsky, M. E., J. K. Shoemaker, and R. L. Hughson. Beat-by-beat forearm blood flow with Doppler ultrasound and strain-gauge plethysmography. J. Appl. Physiol. 79:713-719, 1995.
16. Van Leeuwen, B. E., J. Lubbers, G. J. Barendsen, and L. De PATER. Calf blood flow and posture: Doppler ultrasound calibrated by plethysmography. J. Appl. Physiol. 72:1668-1674, 1992.
17. Walloe, L. and J. Wesche. Time course and magnitude of blood flow changes in the human quadriceps muscles during and following rhythmic exercise. J. Physiol. 405:257-273, 1988.
18. Zananiri, F. V., P. C. Jackson, M. Halliwell, et al. A comparative study of velocity measurements in major blood vessels using magnetic resonance imaging and Doppler ultrasound. Br. J. Radiol. 66:1128-1133, 1993.


©1996The American College of Sports Medicine