- Email: [email protected]
Magnetic resonance measurement of blood flow in peripheral vessels after acute exercise
Mqnetrc Resonance Imaging. Vol. I I, PP. 1085-1092, 1993 Printed in the USA. All rights rexned.
Copyright 0
0730-725X193 $6.00 + .I0 1993 Pergamon Press Ltd.
0 Original Contribution
MAGNETIC RESONANCE MEASUREMENT OF BLOOD FLOW IN PERIPHERAL VESSELS AFTER ACUTE EXERCISE RONALD
A. MEYER,*~
JEANNE M. FOLEY,~$ SUSAN
J. HARKEMA,~
ARLENE SIERRA,* AND E. JAMES POTCHEN* Departments
of *Radiology, tPhysiology, and IHealth and Physical Education, Michigan State University, East Lansing, MI 48824, USA
Velocity-encoded Cine magnetic resonance imaging (MRI) was used to measure blood flow in the anterior tibia1 artery (AT), posterior tibia1 artery (PT), and popliteal artery of adult human subjects (mean age 29 yr) before and after 90 s of ankle dorsiflexion exercise. Before exercise, mean flow, peak systolic velocity, and end-diastolic velocity in AT were 8.1 f 1.6 (SE, n = 6) ml/mitt, 26.9 + 2.6 cm/s, and -0.6 + 0.4 cm/s, respectively. After exercise, mean flow and peak systolic velocity in AT increased by 19-fold and 3-fold, respectively, and end-diastolic velocity increased to 8.7 + 1.1 cm/s. Flow in popliteal artery above its bifurcation was similar to the sum of flows in AT and PT, both before and after exercise. Flow in AT declined exponentially after exercise with a mean half-time of 4 min. The resullts demonstrate the utility of MR phase-encoded flow-velocity measurements for physiological studies of peripheral vascular dynamics after exercise. Keywords:
Magnetic resonance angiography;
Cine phase-contrast:
INTFLODUCTION
Muscleblood flow.
measurements are typically made in large, relatively accessible vessels, for example, the femoral vessels. In this study we applied Cine phase-contrast velocity-encoded magnetic resonance imaging (MRI) to simultaneously measure blood flow in the tibia1 and popliteal arteries of human subjects before and after brief exercise of the ankle flexor muscles. The results demonstrate that MRI is uniquely well-suited for hemodynamic measurements in small peripheral vessels after exercise.
Although there is general agreement that cardiac output can become a limiting factor for performance during whole body exercise in humans,’ it is not known to what extent blood flow limits performance during exercise of relatively small, localized muscle groups in normal or diseased human subjects. A major obstacle to investigating this issue has been lack of a method for measuring flow to specific muscles, for example, flexor vs. extensor muscles. Most published estimates of muscle blood flow in human subjects during and after exercise are based on measurements of total limb flow, rather than on measurelments of flow to specific muscles or muscle groups. The three methods most commonly employed by exercise physiologists for measuring peripheral flow are venous-occlusion plethysmography,’ indicator-dilution catheterization3 and more recently, Doppler ultrasound.4 These methods measure flow to the entire limb distal to the occlusion, to the catheter tip, or to the ultrasound probe, respectively. Although in principle it might be possible to measure flow by catheterization or ultrasound in smaller arteries supplying more limited portions of a limb, in practice these
METHODS A total of eight young adult subjects (Table 1) were recruited from the academic community. The project was approved by the university committee on research involving human subjects, and the subjects gave informed written consent. A4RI Methods Subjects were positioned supine in a 1.5 T Signa whole body imager (General Electric, Milwaukee, WI) with their lower right leg in a custom-made MR-compatible Plexiglas exercise boot equipped with strain
of Physiology, Giltner Hall, Michigan State University, Lansing, MI 48824.
RECEIVED 2/ l/93; ACCEPTED 6/ 15/93. Address correspondence to Ronald A. Meyer, PhD, Dept. 1085
East
Magnetic Resonance Imaging 0 Volume 11, Number 8, 1993
1086 Table 1. Characteristics Sex Age (yr) Weight (kg) Height (cm) Blood pressure (Torr) Systolic Diastolic Heart rate (resting, bpm) Cross-sectional area (cm2) Leg Ant. tibialis
of subjects 4 male, 4 female 29 ? 3 62 ? 3 165 + 2 120 * 3 81 +5 58 + 6 90.1 + 3.2 (n = 6) 9.3 + 0.8 (n = 6)
Values are means f SE.
gauges for measuring the force of ankle dorsiflexion. All images were acquired from a standard linear transceiver coil (GE extremity coil). A set of 2D gradientrecalled-echo time-of-flight flow images’ were acquired to identify suitable axial planes for the flow measurements described below. This localizer set consisted of 48-64 adjacent axial slices (each 2 mm thick, 16 cm FOV, TR/TE = 33/8,60” pulse width, 256 x 128 acquisition matrix, 1 NEX) centered on a region approximately 5 cm below the inferior margin of the patella. Twenty lateral maximum intensity projections were computed from this set of axial images, which taken together provide a 3D representation of the vessels in the imaged region (see Fig. 1). Based on these images, a single axial slice was chosen for the main study which transected both the anterior and posterior tibia1 arteries distal to their bifurcation from the popliteal artery, but superior to the tibial-peroneal bifurcation, and in which the axes of both arteries were within 20” of a line perpendicular to the slice plane. In one of the subjects, the peroneal and posterior tibia1 arteries branched very close to the popliteal bifurcation. In this case flow in peroneal and posterior tibia1 arteries was summed. In a subsequent study, flow images in six subjects were simultaneously obtained from two slices: one transecting the popliteal artery l-2 cm superior to its bifurcation into anterior and posterior tibia1 arteries, and the second below that bifurcation as described above. Four subjects took part in both studies. Images of flow velocity in the selected slices were acquired in retrospectively ECG-gated Cine mode,6*7 as first developed by Pelt et al.’ In brief, the method depends on measurement of the extra phase, 8 = yA TV, acquired by spins moving at velocity, V(cm/s), along the axis of a bipolar flow-encoding gradient, where y (radian/Gauss) is the gyromagnetic ratio, A is the time integral of gradient strength (Gauss*s/cm) and Tis the time between the two gradient lobes.*s9The extra phase due to motion is measured relative to the phase of a
Fig. 1. Maximum pixel intensity projection through a set of 64 axial gradient-recalled-echo images taken below the knee of a subject before exercise. This image is from a set of 20 lateral projections used to identify axial slices suitable for flow measurements (see methods). Note popliteal (POP), anterior tibia1 (AT), and posterior tibia1 (PT) arteries.
flow-compensated scan acquired from the same location with no flow-encoding gradient. In this study, the flow-encoding gradient was applied along the superiorinferior axis (i.e., parallel to the vessels), with maximum velocity sensitivity (VENC, corresponding to ? 180” phase shifts) set at 100 cm/s. Imaging parameters were TRITE = 2618, 1 cm slice thickness, 12-14 cm fieldof-view, 30” pulse, 256 x 128 acquisition matrix, and 1 NEX. Data was collected in Cine mode over 128 heart beats. Assuming a typical beat interval of 1 s (i.e., heart rate [HR] = 60 bpm), these image parameters allow 38 acquisitions (lOOO/TR) per beat, yielding sufficient data after 128 heart beats to reconstruct 19 velocity images gated to regular intervals during the cardiac cycle. In this study we computed 32 images over the cardiac cycle, of which somewhat less than half are, therefore, interpolated, depending on the actual heart rate. If only one slice was imaged, these 32 images corresponded to 32 regularly spaced intervals during the cardiac cycle. If two slices were imaged, images at 16 regularly spaced intervals were computed for each slice. In either case, total acquisition time per set of 32 velocity images was 128 x (l/HR), or about 128 s. Standard MR spin magnitude images reconstructed from the same data were also used to measure cross-sectional areas of the anterior tibialis muscle and of the lower leg (Table 1).
MR angiography after acute exercise 0 R.A. MEYERET
Instantaneous flow (ml/min) was calculated from the individual velocity images by integrating velocity (cm/s) across the area (cm2) of the vessel. Demarcation of vessel pixels w,as performed on a Sun 4 computer as follows. First, the image from each set of 16 or 32 images with the highest peak velocity was identified. Using this image., the mean and standard deviation of velocity of background tissue (muscle and skin) was computed over a large rectangular region of interest which included no visible vessels. A single pixel was then marked at an arbitrary location within each visible vessel in this image by a human observer. The computer then automatically grew a region of all adjacent pixels around each seed pixel in which the velocity was more than 2.5 standard deviations above (arteries) or below (veins) the mean background velocity. These computer expanded ves#selregions were used for instantaneous flow integration in each of the 16 or 32 images in the set. Finally, mean flow in each vessel region was computed from the mean across all 16 or 32 images in the set. Exercise Protocol The subject rested supine in the magnet for 25-30 min before the first, preexercise, flow scan was acquired. Within 5 min after the preexercise scan, the subject performed resisted ankle dorsiflexion (90 s at 1 Hz, 6 kg peak force, or 30% of the mean maximum voluntary force in these subjects). Dorsiflexion was from approximately 130 to 90”, and1was performed against a flexible rubber hose, which served to passively reextend the foot between contractions. This exercise was previously shown by MRI to result in extensive recruitment of anterior tibialis muscle, with little detectable recruitment of posterior calf muscles. lo Heart rate increased during the exercise by lo-Z!0 bpm in all subjects. After the exercise, heart rate returned to resting level after 3060 s, at which point the first postexercise flow scan was
Table 2. Hemodynamic
changes
1087
AL.
initiated. Thus, the first postexercise scan represents events between 1 and 3 min after the exercise. In the initial study, additional flow scans were acquired at 6 min intervals for up to 30 min after the exercise. RESULTS
Cardiac-Gated Velocity and Flow Measurements Sample velocity images from one subject acquired before and after the dorsiflexion exercise appear in Fig. 2. These systolic and end-diastolic images were selected from complete sets of 32 images corresponding to 32 equally spaced intervals during the cardiac cycle before or after the exercise. White in the images is superior-to-inferior motion, black is inferior-to-superior motion and gray is stationary tissue. Regions with very low MR signal intensity, for example, outside the leg or in the matrix of bones, have random phase, and therefore appear as noise. The striking feature of these images is the dramatic increase in flow velocity in the anterior tibia1 artery after the exercise, particularly during diastole. Also noteworthy is the appearance of both the anterior and posterior tibia1 veins after exercise. These vessels were not clearly resolved before exercise. Figure 3 shows peak velocities (typically the velocity in the central pixel of a vessel) during the cardiac cycle in the anterior and posterior tibia1 arteries of a subject before and after the dorsiflexion exercise. Before exercise the velocity waveforms were nearly identical in these two arteries. After the exercise, there was a 3-fold increase in systolic peak velocity, and a much greater increase in end-diastolic peak velocity in the anterior compared to the posterior artery. Similar results were obtained in all subjects (Table 2). The mean background velocity of stationary tissue in the images was -0.02 f 0.02 cm/s (SE, n = 12). The mean of the standard deviations for these background measurements was 2.3 + 0.2 cm/s. Thus, the thresh-
in ant. and post. tibia1 arteries
after ankle dorsiflexion
Anterior PrePeak Systolic Velocity (cm/set) End Diastolic Velocity (cm/set) Mean Velocity (cm/set) Cross-sectional area (cm2) Mean Flow (mUmin)
exercise
Posterior Post-exercise
Pre-
Post-exercise
34.8 f 2.8
35.2 ? 2.9
. 26.9 f 2.6 -0.6 1.4 0.10 8.1
+ + + +
0.4 0.1 .0.02 1.6
Values are means + SE, n = 6. *Significant difference after exercise by paired Student’s
82.1 f 3.1* 8.7 17.5 0.15 153.5
+ f + f
l.l* 1.1* 0.02* 19.1*
t-test, p < .05.
-0.4 1.9 0.23 26.4
+ + * *
0.3 0.3 0.02 5.8
0.0 3.4 0.23 45.9
?I + + +
0.4 0.4* 0.02 7.6*
1088
Magnetic
Resonance
Imaging
(A)
0 Volume
11. Number
8, 1993
(B)
(D) Fig. 2. ECG-gated tine velocity images from one subject acquired at systole (A,C) and end-diastole (B,D) immediately before (A,B) and after (C,D) dorsiflexion exercise. Selected from sets of 32 velocity images corresponding to 33 ms intervals during the cardiac cycle. Note increased flow velocity in anterior tibia1 artery (AT) compared to posterior artery (PT), and appearance of venous flow (dark regions adjacent to arteries), after exercise.
old used for demarcation of arteries in the peak systolic images was typically -0.02 + (2.5 * 2.3), or 5.6 cm/s. Figure 4 shows velocity profiles across the marked region of the posterior tibia1 artery at various times in the cardiac cycle in a subject at rest. Inasmuch as the profiles extrapolate to zero within one pixel distance (0.47 mm) outside the marked region, these profiles show that the demarcation procedure includes all
but a small fraction of the total flow in a vessel. As expetted from simulations of flow in larger arteries,” the velocity profile near peak systole was somewhat flattened compared to the parabolic profile of classic laminar flow. Instantaneous flow at various times during the cardisc cycle can be estimated from the velocity images by integrating velocity across the area of a vessel. In-
MR angiography 90 70
T
after acute exercise 0 R.A. MEYER ET
1089
AL
Pre-exercise
Pre-exercise
O-0
A. Tib.
O---O
P. Tib.
O-0
A. Tib.
0-O
P. Tib.
-30 -50 -70 0
200
I 600
400
I 600
; 1000
400
..a,.* ./
‘. \
50 --
/ 30 -;,a..-.-., .-*._*.... . / IO-_&*.E,._! _ -lO--
t
1000
(A)
(A) go-
800
Time (msec)
Time (msec)
70 --
600
Post-exercise l -e n -m A--* v-v
A. Tib. P. Tib. A. vein P. vein
‘0 ‘..* _**_*../~..*.
Post-exercise
.,*;
350 _I
t
m-0 ?
1
250
. ...._..* 150
.. .
- - - -xi..
A. Tib.
-m
P. Tib.
AL-A
A. vein
V-T
P. vein
n
. -50
.~e?“~-_Em-~=_a. &
-30 -50
t ~-._._._._r_~.._._r_..~~~~T-~.~....~.,..-~...~~~.............~
,....-.-.-.-.-....r.....~.....,-.-.-.-.-.-.-.-.-.........-.-... ._._.-A.A.A-A.A.._;_..*.
0
..A
._._._._*........-A-A-A.~-A-A-A.A
-150 600
400
200
600
1000
0
Time (msec)
200
400
600
600
1000
Time (msec)
(W
(B)
Fig. 3. Peak velocity vs. time during the cardiac cycle in anterior (circles) and posterior (squares) tibia1 arteries before (A) and after (B) ankle dorsiflexion exercise in the same subject as Fig. 2. Zero time corresponds to QRS wave of ECG signal. Also shown in bottom panel are velocities in anterior and posterior tibia1 veins which were resolved in this subject after exercise.
Fig. 5. Flow vs. time during the cardiac cycle in the anterior (circles) and posterior (squares) tibia1 arteries of a subject before (A) and after (B) ankle dorsiflexion exercise in the same subject as Figs. 2 and 3. Also shown in the bottom panel (triangles) are flows in the anterior and posterior tibia1 veins after exercise. Mean flows across the cardiac cycle after the exercise were 142.9, 41.6, -116.3, and -82.7 ml/min, for anterior artery, posterior artery, anterior vein, and posterior vein, respectively.
40-r
-201 -3.0
I -2.0
-1.0
0.0 Distance
1.0
2.0
3.0
(mm)
Fig. 4. Veiocity profiles across the posterior tibia1 artery at various times during the cardiac cycle in a subject before exercise. Times (in ms) were 1198(0), 264 (O), 297 (A),363 (A), 528 (O), 594 (W), and 726 (0) ms after QRS wave. The profiles are across the region marked as a vessel by the procedure described in methods.
stantaneous flows in the vessels of a subject before and after the dorsiflexion exercise appear in Fig. 5. Before exercise, the amplitude of the flow waveform was larger in the posterior than anterior tibia1 artery, reflecting the fact that the posterior artery is larger these subjects (Table 2). There was a clear period of retrograde flow in both arteries, and end-diastolic flow was not significantly different than zero. After exercise, instantaneous flow in the anterior artery increased at all times during the cardiac cycle, and the retrograde phase was abolished. Figure 5 also shows the nonpulsatile flow in the anterior and posterior tibia1 veins, which were also resolved in this subject after the exercise, but not before.
Mean Flow Measurements Mean flow in individual vessels before and after exercise can be estimated from the mean across the car-
Magnetic Resonance Imaging 0 Volume 11,Number 8, 1993
1090
disc cycle of the instantaneous flows measured in each vessel. Mean flow in the anterior tibial artery increased by an average of 1Pfold after the exercise compared to before (Table 2). This increase was largely due to increased flow velocity during diastole (e.g, Fig. 3), although there was also a 50% increase in cross-sectional area of the anterior artery after the exercise. Flow also increased by 74% in the posterior tibia1 artery after the exercise, but there was no significant change in area in this vessel (Table 2). Concurrent measurements of flow in several vessels enabled two simple tests of the internal consistency of these measurements. First, in six subjects, flow measurements were simultaneously made in the tibia1 arteries and in the popliteal artery superior to its bifurcation into anterior and posterior tibia1 arteries. Flow in the popliteal artery was within 10% and 20% of the sum in flows in the tibia1 arteries before and after the exercise, respectively (Table 3). The small but significant differences between these measurements could reflect flow in unresolved arterial branches, as well as error in the method. Second, in three of the eight subjects, both anterior and posterior tibia1 veins were resolved after (but not before) the exercise (e.g., Fig. 2). In these subjects, the sum of tibia1 venous flows (171 f 24 ml/min, mean + SE) was not significantly different than the sum of arterial flows (190 f 12 ml/min). After the exercise, flow in the anterior tibia1 arteries of these subjects decreased with a mean half-time of 4.07 + 0.55 (SE, n = 6) min (Fig. 6). This pattern of flow recovery was very reproducible, both on repeated examinations of one subject over a six month period (Fig. 6A), and within this relatively homogeneous group of subjects (Fig. 6B), although the peak flow acheived varied over a 3-fold range between different subjects. DISCUSSION
These results show that MRI has several advantages for measuring peripheral blood flow compared to pre-
Table 3. Comparison of mean flow (ml/min) above and below popliteal bifurcation
Popliteal artery Pre-exercise Post-exercise
55.0 Ifr 8.5 199 + 37
Ant. plus post. tibia1 arteries 50.5 f s.9* 163 + 30*
Values are means f SE, n = 6. *Significantly different from popliteal flow by paired Student’s t-test, p < .05.
150 125
T
01 -10
I
I
I
0
10
30
20
Time (min) (A)
160
60 40 0 -10
_ 0
10
20
;0
Time (min) (R) Fig. 6. Changes in mean flow in arteries during recovery after ankle dorsiflexion exercise. (A) Results in anterior (filled symbols) and posterior (open symbols) arteries from one subject examined on three occasions over a 6-mo period. (B) Results in anterior artery from 6 subjects.
viously available methods. Unlike thermodilution or other tracer techniques, MRI is completely noninvasive, and has no known adverse effects. Of the other available methods, Doppler ultrasound’2,‘3 is the most analogous to MRI, insofar as both methods measure the vector component of flow along some axis. However, the tissue penetration and spatial resolution of MRI are clearly superior to those of ultrasound. Moreover, the field-of-view in Doppler ultrasound flow imaging is restricted to a relatively narrow angle projecting from the head of the ultrasound probe. In contrast, MRI offers the unique advantages that quantitative flow measurements can be made along any axis,’ and can be simultaneously made in many widely separated vessels or in different locations along the same vessel. In this study, the simultaneous imaging of flow in several vessels enabled two checks on the internal consistency of the measurements. First, the sum of arterial flows below the popliteal bifurcation was similar
MR angiography
after acute exercise 0 R.A. MEYER ET AL.
to the flow in the pophteal artery. Although there was slightly more flow in lthe popliteal, particularly after the exercise, this was not surprising, because there are probably small, unresolved arteries branching to the portion of the anterior muscles superior to the lower slice. Second, in subjects in which both anterior and posterior tibia1 veins were resolved, the sum of venous flows equaled the sum of arterial flows through the same slice. This check is particularly reassuring because the velocity waveforms are highly pulsatile in the arteries but not in the veins. For example, it has been shown that the interpolated reconstruction method used in this study suppresses high frequency components of flow waveforms. l4 While this and other effects of pulsatility may produce errors in measurements of peak flow, it appears from our results that they do not seriously compromise mean flow measurements. One disadvantage that MRI shares with ultrasound and indicator-dilution methods is that it measures vessel blood flow rather than true muscle perfusion (i.e., ml blood/min/ml muscle). Because it is the latter which is primarily of interest in exercise studies, with all of these methods it is necessary to independently estimate the muscle mass perfused by the vessel. Furthermore, some correction may be necessary to account for the likelihood that a fraction of the observed flow perfuses other tissues (skin, bones), or other muscles which were not recruited by the exercise. Nonetheless, these problems are simplified by the better spatial resolution of MRI, because the measurements can be made in smaller vessels supplying a maore localized region of tissue. Moreover, standard M[R images can be used to measure the cross-sectional areas or volumes of individual muscles in a limb. In this study, preexercise flow in the popliteal artery subjects was 55 ml/min, which is consistent with previous measurements in this artery by Doppler ultrasound,15 and to estimates of total lower leg flow by venous-occI.usion plethysmography.3 Flow in the anterior tibia1 artery increased from 8 to 153 ml/min, or by 19-fold, after dorsiflexion exercise in these subjects. The volume of anterior tibialis muscle distal to the imaged slice in our subjects was about 100 ml (9 cm2 x 11 cm long cylinder), so this corresponds to an increase from 8 to 153 ml/min/l00 ml muscle. This is comparable to previous measurements of blood flow after maximal exercise in both human’s2 and animal’6”7 muscles. At present, the major technical disadvantage of MRI flow measurements for exercise studies is that they cannot easily be made during dynamic exercise because of movement artifacts in .the images. However, data acquisition could in principle be gated to rhythmic limb movements as well as to the cardiac cycle. Finally, the relatively long time (2 min) needed to acquire a full set
1091
of gated images is a potential disadvantage, particularly when flow is changing, as during the recovery period in this study. However, these flow measurements are not a simple linear average of flow over the acquisition time, because most of the acquisition time is expended acquiring high spatial frequency information, which makes a relatively minor contribution to the total signal, at least in larger vessels. Therefore, it may be possible to shorten the imaging time to 30 s or less by acquiring fewer phase-encode steps, without seriously compromising the flow measurements. In any case, this time limitation may soon be overcome by use of echoplanar and other fast MR imaging techniques.” One potential new application suggested by our results is the possibility of estimating the extent of collateralization between peripheral vessels in human subjects. In particular, venous collateralization was clearly more extensive in some subjects (e.g. Figs. 2 and 5), inasmuch as posterior venous flow was much greater than posterior artery flow after the exercise. In all subjects there was also a significant increase in posterior as well as anterior artery flow after the exercise, despite the fact that dorsiflexion exercise nominally recruits only anterior compartment muscle.” However, this increase might reflect increased perfusion of the posterior muscles, as well as collateral flow to anterior muscle. Our results also suggest that a brief bout of acute exercise could be used to enhance vessel contrast during diagnostic MR studies of the peripheral vasculature. First, venous structures which are not resolved at rest can be easily distinguished after exercise. Furthermore, the elimination of the retrograde phase in arterial flow after exercise should considerably enhance arterial signal during ungated studies, especially when inferior saturation is applied to eliminate venous flow. In a previous studylo using the same exercise protocol and a similar subject population as in this study, we reported that the increased muscle T2 that occurs after acute exercise recovers with a half-time over 10 min. In contrast, blood flow in this study recovered after exercise with a half-time of 4 min. Thus, as recently concluded by Fleckenstein and co-workers,” the T2 change observed by MRI after acute exercise is not directly related to changes in muscle perfusion. In summary, this study demonstrates that reproducible and internally consistent measurements of exerciseinduced changes in peripheral blood flow are possible by phase-contrast velocity-encoded MRI. The method should prove useful for quantitative peripheral flow studies in both normal and diseased human subjects. AcknowledgmentsWe gratefully acknowledge the assistance of James Seibert and Thomas Cooper, Dept. of Radiology, Michigan State University. Supported in part by NIH AR38972.
092
Magnetic Resonance Imaging 0 Volume 11, Number 8, 1993
REFERENCES 1. Rowell, L.B. Muscle blood flow in humans: 2.
3.
4.
5. 6.
7.
8.
9.
How high can it go? Med. Sci. Sports Exert. 2O:S97-S103; 1988. Martin, W.H., III; Kohrt W.H.; Malley W.M.; Korte M.T.; Stoltz, S. Exercise training enhances leg vasodilatory capacity of 6%year-old men and women. J. Appl. Physiol. 69:1804-1809; 1990. Sullivan, M. J.; Binkley, P.K.; Unverferth, D.V.; Leier, C.V. Hemodynamic and metabolic responses of the exercising lower limb of humans. J. Lab. C/in. Med. 110: 145-152; 1987. Vollestad, N.K.; Wesche, J.; Sejersted, O.M. Gradual increase in leg oxygen uptake during repeated submaximal contractions in humans. J. Appl. Physiol. 68: 1150-l 156; 1990. Nishimura, D.G. Time-of-flight MR angiography. Mugn. Reson. Med. 14:194-201; 1990. Caputo, G.R.; Masui,T.;Gooding, G.A.W.;Chang, J.M.; Higgins, C.B. Popliteal and tibioperoneal arteries: Feasibility of two-dimensional time-of-flight MR angiography and phase velocity mapping. Radiology 182: 387-392; 1992. Caputo, G.R.; Kondo, C.; Masui, F.; Geraci, S. J.; Foster, E.; O’Sullivan, M.M.; Higgins, C.B. Right and left lung perfusion: In vitro and in vivo validation with oblique-angle, velocity-encoded Cine MR imaging. Radiology 180:693-698; 1991. Pelt, N.J.; Shimakawa, A.; Glover, G.H. Phase-contrast tine MRI. In: Book of abstracts: Eighth Annual Meeting of the Society of Magnetic Resonance in Medicine. Berkeley, CA:SMRM; 1989: 101. Walker, M.F.; Souza, S.P.; Dumoulin, C.L. Quantitative flow measurement in phase contrast MR angiography. J. Comp. Assist. Tomogr. 12:304-313; 1988.
10. Fisher, M.J.; Meyer, R.A; Adams, G.R.; Foley, J.M.; Potchen, E.J. Direct relationship between proton T, and exercise intensity in skeletal muscle MR images. fnvest. Radiol. 25:480-485; 1990. 1 I. Perktold, K.; Hilbert, D. Numerical simulation of pulsatile flow in a carotid bifurcation model. J. Biomed. Eng. 8:193-199; 1986. 12. Grant, E.G.; Tessler, R.N.; Perrella, R.R. Clinical doppler imaging. Am. J. Radio/. 152:707-717; 1989. 13. Greene, E.R. Noninvasive measurement of blood flow using pulsed Doppler ultrasound. In: J.A. Loeppky, M.L. Riedesel (Eds). Oxygen Transport to Human Tissues. New York: Elsevier/North Holland; 1982: pp. 135-150. 14. Frayne, R.; Rutt, B.K. Frequency response of Cine phase contrast. In: Book of abstracts: Eleventh Annual Meeting of the Society of Magnetic Resonance in Medicine. Berkeley, CA:SMRM; 1992: 2907. 15. Fronek, A.; Coel, M.; Bernstein, E.F. Quantitative ultrasonographic studies of lower extremity flow velocities in health and disease. Circulation 53:957-960; 1976. 16. Mackie, B.G.; Terjung, R.L. The influence of training on blood flow to the different fiber types in the rat skeletal muscle. Am. J. Physiol. 237:Cll l-Cl 18; 1983. 17. Petrofsky, J.S.; Phillips, C.A.; Sawka, M.N.; Hanpeter, D.; Stafford, D. Blood flow and metabolism during isometric contractions in cat skeletal muscle. J. Appl. Physioi. 50:493-502; 1981. 18. Guilfoyle, D.N.; Gibbs, P.; Ordidge, R. J.; Mansfield, P. Real-time flow measurements using echo-planar imaging. Magn. Reson. Med. 18:1-8; 1991. 19. Fleckenstein, J.L.; Haller, R.G.; Bertocci, L.A.; Parkey, R.W.; Peshock, R.M. Glycogenolysis, not perfusion, is the critical mediator of exercise-induced muscle modifications on MR images. Radiology 183:25-27; 1992.
Copyright 0
0730-725X193 $6.00 + .I0 1993 Pergamon Press Ltd.
0 Original Contribution
MAGNETIC RESONANCE MEASUREMENT OF BLOOD FLOW IN PERIPHERAL VESSELS AFTER ACUTE EXERCISE RONALD
A. MEYER,*~
JEANNE M. FOLEY,~$ SUSAN
J. HARKEMA,~
ARLENE SIERRA,* AND E. JAMES POTCHEN* Departments
of *Radiology, tPhysiology, and IHealth and Physical Education, Michigan State University, East Lansing, MI 48824, USA
Velocity-encoded Cine magnetic resonance imaging (MRI) was used to measure blood flow in the anterior tibia1 artery (AT), posterior tibia1 artery (PT), and popliteal artery of adult human subjects (mean age 29 yr) before and after 90 s of ankle dorsiflexion exercise. Before exercise, mean flow, peak systolic velocity, and end-diastolic velocity in AT were 8.1 f 1.6 (SE, n = 6) ml/mitt, 26.9 + 2.6 cm/s, and -0.6 + 0.4 cm/s, respectively. After exercise, mean flow and peak systolic velocity in AT increased by 19-fold and 3-fold, respectively, and end-diastolic velocity increased to 8.7 + 1.1 cm/s. Flow in popliteal artery above its bifurcation was similar to the sum of flows in AT and PT, both before and after exercise. Flow in AT declined exponentially after exercise with a mean half-time of 4 min. The resullts demonstrate the utility of MR phase-encoded flow-velocity measurements for physiological studies of peripheral vascular dynamics after exercise. Keywords:
Magnetic resonance angiography;
Cine phase-contrast:
INTFLODUCTION
Muscleblood flow.
measurements are typically made in large, relatively accessible vessels, for example, the femoral vessels. In this study we applied Cine phase-contrast velocity-encoded magnetic resonance imaging (MRI) to simultaneously measure blood flow in the tibia1 and popliteal arteries of human subjects before and after brief exercise of the ankle flexor muscles. The results demonstrate that MRI is uniquely well-suited for hemodynamic measurements in small peripheral vessels after exercise.
Although there is general agreement that cardiac output can become a limiting factor for performance during whole body exercise in humans,’ it is not known to what extent blood flow limits performance during exercise of relatively small, localized muscle groups in normal or diseased human subjects. A major obstacle to investigating this issue has been lack of a method for measuring flow to specific muscles, for example, flexor vs. extensor muscles. Most published estimates of muscle blood flow in human subjects during and after exercise are based on measurements of total limb flow, rather than on measurelments of flow to specific muscles or muscle groups. The three methods most commonly employed by exercise physiologists for measuring peripheral flow are venous-occlusion plethysmography,’ indicator-dilution catheterization3 and more recently, Doppler ultrasound.4 These methods measure flow to the entire limb distal to the occlusion, to the catheter tip, or to the ultrasound probe, respectively. Although in principle it might be possible to measure flow by catheterization or ultrasound in smaller arteries supplying more limited portions of a limb, in practice these
METHODS A total of eight young adult subjects (Table 1) were recruited from the academic community. The project was approved by the university committee on research involving human subjects, and the subjects gave informed written consent. A4RI Methods Subjects were positioned supine in a 1.5 T Signa whole body imager (General Electric, Milwaukee, WI) with their lower right leg in a custom-made MR-compatible Plexiglas exercise boot equipped with strain
of Physiology, Giltner Hall, Michigan State University, Lansing, MI 48824.
RECEIVED 2/ l/93; ACCEPTED 6/ 15/93. Address correspondence to Ronald A. Meyer, PhD, Dept. 1085
East
Magnetic Resonance Imaging 0 Volume 11, Number 8, 1993
1086 Table 1. Characteristics Sex Age (yr) Weight (kg) Height (cm) Blood pressure (Torr) Systolic Diastolic Heart rate (resting, bpm) Cross-sectional area (cm2) Leg Ant. tibialis
of subjects 4 male, 4 female 29 ? 3 62 ? 3 165 + 2 120 * 3 81 +5 58 + 6 90.1 + 3.2 (n = 6) 9.3 + 0.8 (n = 6)
Values are means f SE.
gauges for measuring the force of ankle dorsiflexion. All images were acquired from a standard linear transceiver coil (GE extremity coil). A set of 2D gradientrecalled-echo time-of-flight flow images’ were acquired to identify suitable axial planes for the flow measurements described below. This localizer set consisted of 48-64 adjacent axial slices (each 2 mm thick, 16 cm FOV, TR/TE = 33/8,60” pulse width, 256 x 128 acquisition matrix, 1 NEX) centered on a region approximately 5 cm below the inferior margin of the patella. Twenty lateral maximum intensity projections were computed from this set of axial images, which taken together provide a 3D representation of the vessels in the imaged region (see Fig. 1). Based on these images, a single axial slice was chosen for the main study which transected both the anterior and posterior tibia1 arteries distal to their bifurcation from the popliteal artery, but superior to the tibial-peroneal bifurcation, and in which the axes of both arteries were within 20” of a line perpendicular to the slice plane. In one of the subjects, the peroneal and posterior tibia1 arteries branched very close to the popliteal bifurcation. In this case flow in peroneal and posterior tibia1 arteries was summed. In a subsequent study, flow images in six subjects were simultaneously obtained from two slices: one transecting the popliteal artery l-2 cm superior to its bifurcation into anterior and posterior tibia1 arteries, and the second below that bifurcation as described above. Four subjects took part in both studies. Images of flow velocity in the selected slices were acquired in retrospectively ECG-gated Cine mode,6*7 as first developed by Pelt et al.’ In brief, the method depends on measurement of the extra phase, 8 = yA TV, acquired by spins moving at velocity, V(cm/s), along the axis of a bipolar flow-encoding gradient, where y (radian/Gauss) is the gyromagnetic ratio, A is the time integral of gradient strength (Gauss*s/cm) and Tis the time between the two gradient lobes.*s9The extra phase due to motion is measured relative to the phase of a
Fig. 1. Maximum pixel intensity projection through a set of 64 axial gradient-recalled-echo images taken below the knee of a subject before exercise. This image is from a set of 20 lateral projections used to identify axial slices suitable for flow measurements (see methods). Note popliteal (POP), anterior tibia1 (AT), and posterior tibia1 (PT) arteries.
flow-compensated scan acquired from the same location with no flow-encoding gradient. In this study, the flow-encoding gradient was applied along the superiorinferior axis (i.e., parallel to the vessels), with maximum velocity sensitivity (VENC, corresponding to ? 180” phase shifts) set at 100 cm/s. Imaging parameters were TRITE = 2618, 1 cm slice thickness, 12-14 cm fieldof-view, 30” pulse, 256 x 128 acquisition matrix, and 1 NEX. Data was collected in Cine mode over 128 heart beats. Assuming a typical beat interval of 1 s (i.e., heart rate [HR] = 60 bpm), these image parameters allow 38 acquisitions (lOOO/TR) per beat, yielding sufficient data after 128 heart beats to reconstruct 19 velocity images gated to regular intervals during the cardiac cycle. In this study we computed 32 images over the cardiac cycle, of which somewhat less than half are, therefore, interpolated, depending on the actual heart rate. If only one slice was imaged, these 32 images corresponded to 32 regularly spaced intervals during the cardiac cycle. If two slices were imaged, images at 16 regularly spaced intervals were computed for each slice. In either case, total acquisition time per set of 32 velocity images was 128 x (l/HR), or about 128 s. Standard MR spin magnitude images reconstructed from the same data were also used to measure cross-sectional areas of the anterior tibialis muscle and of the lower leg (Table 1).
MR angiography after acute exercise 0 R.A. MEYERET
Instantaneous flow (ml/min) was calculated from the individual velocity images by integrating velocity (cm/s) across the area (cm2) of the vessel. Demarcation of vessel pixels w,as performed on a Sun 4 computer as follows. First, the image from each set of 16 or 32 images with the highest peak velocity was identified. Using this image., the mean and standard deviation of velocity of background tissue (muscle and skin) was computed over a large rectangular region of interest which included no visible vessels. A single pixel was then marked at an arbitrary location within each visible vessel in this image by a human observer. The computer then automatically grew a region of all adjacent pixels around each seed pixel in which the velocity was more than 2.5 standard deviations above (arteries) or below (veins) the mean background velocity. These computer expanded ves#selregions were used for instantaneous flow integration in each of the 16 or 32 images in the set. Finally, mean flow in each vessel region was computed from the mean across all 16 or 32 images in the set. Exercise Protocol The subject rested supine in the magnet for 25-30 min before the first, preexercise, flow scan was acquired. Within 5 min after the preexercise scan, the subject performed resisted ankle dorsiflexion (90 s at 1 Hz, 6 kg peak force, or 30% of the mean maximum voluntary force in these subjects). Dorsiflexion was from approximately 130 to 90”, and1was performed against a flexible rubber hose, which served to passively reextend the foot between contractions. This exercise was previously shown by MRI to result in extensive recruitment of anterior tibialis muscle, with little detectable recruitment of posterior calf muscles. lo Heart rate increased during the exercise by lo-Z!0 bpm in all subjects. After the exercise, heart rate returned to resting level after 3060 s, at which point the first postexercise flow scan was
Table 2. Hemodynamic
changes
1087
AL.
initiated. Thus, the first postexercise scan represents events between 1 and 3 min after the exercise. In the initial study, additional flow scans were acquired at 6 min intervals for up to 30 min after the exercise. RESULTS
Cardiac-Gated Velocity and Flow Measurements Sample velocity images from one subject acquired before and after the dorsiflexion exercise appear in Fig. 2. These systolic and end-diastolic images were selected from complete sets of 32 images corresponding to 32 equally spaced intervals during the cardiac cycle before or after the exercise. White in the images is superior-to-inferior motion, black is inferior-to-superior motion and gray is stationary tissue. Regions with very low MR signal intensity, for example, outside the leg or in the matrix of bones, have random phase, and therefore appear as noise. The striking feature of these images is the dramatic increase in flow velocity in the anterior tibia1 artery after the exercise, particularly during diastole. Also noteworthy is the appearance of both the anterior and posterior tibia1 veins after exercise. These vessels were not clearly resolved before exercise. Figure 3 shows peak velocities (typically the velocity in the central pixel of a vessel) during the cardiac cycle in the anterior and posterior tibia1 arteries of a subject before and after the dorsiflexion exercise. Before exercise the velocity waveforms were nearly identical in these two arteries. After the exercise, there was a 3-fold increase in systolic peak velocity, and a much greater increase in end-diastolic peak velocity in the anterior compared to the posterior artery. Similar results were obtained in all subjects (Table 2). The mean background velocity of stationary tissue in the images was -0.02 f 0.02 cm/s (SE, n = 12). The mean of the standard deviations for these background measurements was 2.3 + 0.2 cm/s. Thus, the thresh-
in ant. and post. tibia1 arteries
after ankle dorsiflexion
Anterior PrePeak Systolic Velocity (cm/set) End Diastolic Velocity (cm/set) Mean Velocity (cm/set) Cross-sectional area (cm2) Mean Flow (mUmin)
exercise
Posterior Post-exercise
Pre-
Post-exercise
34.8 f 2.8
35.2 ? 2.9
. 26.9 f 2.6 -0.6 1.4 0.10 8.1
+ + + +
0.4 0.1 .0.02 1.6
Values are means + SE, n = 6. *Significant difference after exercise by paired Student’s
82.1 f 3.1* 8.7 17.5 0.15 153.5
+ f + f
l.l* 1.1* 0.02* 19.1*
t-test, p < .05.
-0.4 1.9 0.23 26.4
+ + * *
0.3 0.3 0.02 5.8
0.0 3.4 0.23 45.9
?I + + +
0.4 0.4* 0.02 7.6*
1088
Magnetic
Resonance
Imaging
(A)
0 Volume
11. Number
8, 1993
(B)
(D) Fig. 2. ECG-gated tine velocity images from one subject acquired at systole (A,C) and end-diastole (B,D) immediately before (A,B) and after (C,D) dorsiflexion exercise. Selected from sets of 32 velocity images corresponding to 33 ms intervals during the cardiac cycle. Note increased flow velocity in anterior tibia1 artery (AT) compared to posterior artery (PT), and appearance of venous flow (dark regions adjacent to arteries), after exercise.
old used for demarcation of arteries in the peak systolic images was typically -0.02 + (2.5 * 2.3), or 5.6 cm/s. Figure 4 shows velocity profiles across the marked region of the posterior tibia1 artery at various times in the cardiac cycle in a subject at rest. Inasmuch as the profiles extrapolate to zero within one pixel distance (0.47 mm) outside the marked region, these profiles show that the demarcation procedure includes all
but a small fraction of the total flow in a vessel. As expetted from simulations of flow in larger arteries,” the velocity profile near peak systole was somewhat flattened compared to the parabolic profile of classic laminar flow. Instantaneous flow at various times during the cardisc cycle can be estimated from the velocity images by integrating velocity across the area of a vessel. In-
MR angiography 90 70
T
after acute exercise 0 R.A. MEYER ET
1089
AL
Pre-exercise
Pre-exercise
O-0
A. Tib.
O---O
P. Tib.
O-0
A. Tib.
0-O
P. Tib.
-30 -50 -70 0
200
I 600
400
I 600
; 1000
400
..a,.* ./
‘. \
50 --
/ 30 -;,a..-.-., .-*._*.... . / IO-_&*.E,._! _ -lO--
t
1000
(A)
(A) go-
800
Time (msec)
Time (msec)
70 --
600
Post-exercise l -e n -m A--* v-v
A. Tib. P. Tib. A. vein P. vein
‘0 ‘..* _**_*../~..*.
Post-exercise
.,*;
350 _I
t
m-0 ?
1
250
. ...._..* 150
.. .
- - - -xi..
A. Tib.
-m
P. Tib.
AL-A
A. vein
V-T
P. vein
n
. -50
.~e?“~-_Em-~=_a. &
-30 -50
t ~-._._._._r_~.._._r_..~~~~T-~.~....~.,..-~...~~~.............~
,....-.-.-.-.-....r.....~.....,-.-.-.-.-.-.-.-.-.........-.-... ._._.-A.A.A-A.A.._;_..*.
0
..A
._._._._*........-A-A-A.~-A-A-A.A
-150 600
400
200
600
1000
0
Time (msec)
200
400
600
600
1000
Time (msec)
(W
(B)
Fig. 3. Peak velocity vs. time during the cardiac cycle in anterior (circles) and posterior (squares) tibia1 arteries before (A) and after (B) ankle dorsiflexion exercise in the same subject as Fig. 2. Zero time corresponds to QRS wave of ECG signal. Also shown in bottom panel are velocities in anterior and posterior tibia1 veins which were resolved in this subject after exercise.
Fig. 5. Flow vs. time during the cardiac cycle in the anterior (circles) and posterior (squares) tibia1 arteries of a subject before (A) and after (B) ankle dorsiflexion exercise in the same subject as Figs. 2 and 3. Also shown in the bottom panel (triangles) are flows in the anterior and posterior tibia1 veins after exercise. Mean flows across the cardiac cycle after the exercise were 142.9, 41.6, -116.3, and -82.7 ml/min, for anterior artery, posterior artery, anterior vein, and posterior vein, respectively.
40-r
-201 -3.0
I -2.0
-1.0
0.0 Distance
1.0
2.0
3.0
(mm)
Fig. 4. Veiocity profiles across the posterior tibia1 artery at various times during the cardiac cycle in a subject before exercise. Times (in ms) were 1198(0), 264 (O), 297 (A),363 (A), 528 (O), 594 (W), and 726 (0) ms after QRS wave. The profiles are across the region marked as a vessel by the procedure described in methods.
stantaneous flows in the vessels of a subject before and after the dorsiflexion exercise appear in Fig. 5. Before exercise, the amplitude of the flow waveform was larger in the posterior than anterior tibia1 artery, reflecting the fact that the posterior artery is larger these subjects (Table 2). There was a clear period of retrograde flow in both arteries, and end-diastolic flow was not significantly different than zero. After exercise, instantaneous flow in the anterior artery increased at all times during the cardiac cycle, and the retrograde phase was abolished. Figure 5 also shows the nonpulsatile flow in the anterior and posterior tibia1 veins, which were also resolved in this subject after the exercise, but not before.
Mean Flow Measurements Mean flow in individual vessels before and after exercise can be estimated from the mean across the car-
Magnetic Resonance Imaging 0 Volume 11,Number 8, 1993
1090
disc cycle of the instantaneous flows measured in each vessel. Mean flow in the anterior tibial artery increased by an average of 1Pfold after the exercise compared to before (Table 2). This increase was largely due to increased flow velocity during diastole (e.g, Fig. 3), although there was also a 50% increase in cross-sectional area of the anterior artery after the exercise. Flow also increased by 74% in the posterior tibia1 artery after the exercise, but there was no significant change in area in this vessel (Table 2). Concurrent measurements of flow in several vessels enabled two simple tests of the internal consistency of these measurements. First, in six subjects, flow measurements were simultaneously made in the tibia1 arteries and in the popliteal artery superior to its bifurcation into anterior and posterior tibia1 arteries. Flow in the popliteal artery was within 10% and 20% of the sum in flows in the tibia1 arteries before and after the exercise, respectively (Table 3). The small but significant differences between these measurements could reflect flow in unresolved arterial branches, as well as error in the method. Second, in three of the eight subjects, both anterior and posterior tibia1 veins were resolved after (but not before) the exercise (e.g., Fig. 2). In these subjects, the sum of tibia1 venous flows (171 f 24 ml/min, mean + SE) was not significantly different than the sum of arterial flows (190 f 12 ml/min). After the exercise, flow in the anterior tibia1 arteries of these subjects decreased with a mean half-time of 4.07 + 0.55 (SE, n = 6) min (Fig. 6). This pattern of flow recovery was very reproducible, both on repeated examinations of one subject over a six month period (Fig. 6A), and within this relatively homogeneous group of subjects (Fig. 6B), although the peak flow acheived varied over a 3-fold range between different subjects. DISCUSSION
These results show that MRI has several advantages for measuring peripheral blood flow compared to pre-
Table 3. Comparison of mean flow (ml/min) above and below popliteal bifurcation
Popliteal artery Pre-exercise Post-exercise
55.0 Ifr 8.5 199 + 37
Ant. plus post. tibia1 arteries 50.5 f s.9* 163 + 30*
Values are means f SE, n = 6. *Significantly different from popliteal flow by paired Student’s t-test, p < .05.
150 125
T
01 -10
I
I
I
0
10
30
20
Time (min) (A)
160
60 40 0 -10
_ 0
10
20
;0
Time (min) (R) Fig. 6. Changes in mean flow in arteries during recovery after ankle dorsiflexion exercise. (A) Results in anterior (filled symbols) and posterior (open symbols) arteries from one subject examined on three occasions over a 6-mo period. (B) Results in anterior artery from 6 subjects.
viously available methods. Unlike thermodilution or other tracer techniques, MRI is completely noninvasive, and has no known adverse effects. Of the other available methods, Doppler ultrasound’2,‘3 is the most analogous to MRI, insofar as both methods measure the vector component of flow along some axis. However, the tissue penetration and spatial resolution of MRI are clearly superior to those of ultrasound. Moreover, the field-of-view in Doppler ultrasound flow imaging is restricted to a relatively narrow angle projecting from the head of the ultrasound probe. In contrast, MRI offers the unique advantages that quantitative flow measurements can be made along any axis,’ and can be simultaneously made in many widely separated vessels or in different locations along the same vessel. In this study, the simultaneous imaging of flow in several vessels enabled two checks on the internal consistency of the measurements. First, the sum of arterial flows below the popliteal bifurcation was similar
MR angiography
after acute exercise 0 R.A. MEYER ET AL.
to the flow in the pophteal artery. Although there was slightly more flow in lthe popliteal, particularly after the exercise, this was not surprising, because there are probably small, unresolved arteries branching to the portion of the anterior muscles superior to the lower slice. Second, in subjects in which both anterior and posterior tibia1 veins were resolved, the sum of venous flows equaled the sum of arterial flows through the same slice. This check is particularly reassuring because the velocity waveforms are highly pulsatile in the arteries but not in the veins. For example, it has been shown that the interpolated reconstruction method used in this study suppresses high frequency components of flow waveforms. l4 While this and other effects of pulsatility may produce errors in measurements of peak flow, it appears from our results that they do not seriously compromise mean flow measurements. One disadvantage that MRI shares with ultrasound and indicator-dilution methods is that it measures vessel blood flow rather than true muscle perfusion (i.e., ml blood/min/ml muscle). Because it is the latter which is primarily of interest in exercise studies, with all of these methods it is necessary to independently estimate the muscle mass perfused by the vessel. Furthermore, some correction may be necessary to account for the likelihood that a fraction of the observed flow perfuses other tissues (skin, bones), or other muscles which were not recruited by the exercise. Nonetheless, these problems are simplified by the better spatial resolution of MRI, because the measurements can be made in smaller vessels supplying a maore localized region of tissue. Moreover, standard M[R images can be used to measure the cross-sectional areas or volumes of individual muscles in a limb. In this study, preexercise flow in the popliteal artery subjects was 55 ml/min, which is consistent with previous measurements in this artery by Doppler ultrasound,15 and to estimates of total lower leg flow by venous-occI.usion plethysmography.3 Flow in the anterior tibia1 artery increased from 8 to 153 ml/min, or by 19-fold, after dorsiflexion exercise in these subjects. The volume of anterior tibialis muscle distal to the imaged slice in our subjects was about 100 ml (9 cm2 x 11 cm long cylinder), so this corresponds to an increase from 8 to 153 ml/min/l00 ml muscle. This is comparable to previous measurements of blood flow after maximal exercise in both human’s2 and animal’6”7 muscles. At present, the major technical disadvantage of MRI flow measurements for exercise studies is that they cannot easily be made during dynamic exercise because of movement artifacts in .the images. However, data acquisition could in principle be gated to rhythmic limb movements as well as to the cardiac cycle. Finally, the relatively long time (2 min) needed to acquire a full set
1091
of gated images is a potential disadvantage, particularly when flow is changing, as during the recovery period in this study. However, these flow measurements are not a simple linear average of flow over the acquisition time, because most of the acquisition time is expended acquiring high spatial frequency information, which makes a relatively minor contribution to the total signal, at least in larger vessels. Therefore, it may be possible to shorten the imaging time to 30 s or less by acquiring fewer phase-encode steps, without seriously compromising the flow measurements. In any case, this time limitation may soon be overcome by use of echoplanar and other fast MR imaging techniques.” One potential new application suggested by our results is the possibility of estimating the extent of collateralization between peripheral vessels in human subjects. In particular, venous collateralization was clearly more extensive in some subjects (e.g. Figs. 2 and 5), inasmuch as posterior venous flow was much greater than posterior artery flow after the exercise. In all subjects there was also a significant increase in posterior as well as anterior artery flow after the exercise, despite the fact that dorsiflexion exercise nominally recruits only anterior compartment muscle.” However, this increase might reflect increased perfusion of the posterior muscles, as well as collateral flow to anterior muscle. Our results also suggest that a brief bout of acute exercise could be used to enhance vessel contrast during diagnostic MR studies of the peripheral vasculature. First, venous structures which are not resolved at rest can be easily distinguished after exercise. Furthermore, the elimination of the retrograde phase in arterial flow after exercise should considerably enhance arterial signal during ungated studies, especially when inferior saturation is applied to eliminate venous flow. In a previous studylo using the same exercise protocol and a similar subject population as in this study, we reported that the increased muscle T2 that occurs after acute exercise recovers with a half-time over 10 min. In contrast, blood flow in this study recovered after exercise with a half-time of 4 min. Thus, as recently concluded by Fleckenstein and co-workers,” the T2 change observed by MRI after acute exercise is not directly related to changes in muscle perfusion. In summary, this study demonstrates that reproducible and internally consistent measurements of exerciseinduced changes in peripheral blood flow are possible by phase-contrast velocity-encoded MRI. The method should prove useful for quantitative peripheral flow studies in both normal and diseased human subjects. AcknowledgmentsWe gratefully acknowledge the assistance of James Seibert and Thomas Cooper, Dept. of Radiology, Michigan State University. Supported in part by NIH AR38972.
092
Magnetic Resonance Imaging 0 Volume 11, Number 8, 1993
REFERENCES 1. Rowell, L.B. Muscle blood flow in humans: 2.
3.
4.
5. 6.
7.
8.
9.
How high can it go? Med. Sci. Sports Exert. 2O:S97-S103; 1988. Martin, W.H., III; Kohrt W.H.; Malley W.M.; Korte M.T.; Stoltz, S. Exercise training enhances leg vasodilatory capacity of 6%year-old men and women. J. Appl. Physiol. 69:1804-1809; 1990. Sullivan, M. J.; Binkley, P.K.; Unverferth, D.V.; Leier, C.V. Hemodynamic and metabolic responses of the exercising lower limb of humans. J. Lab. C/in. Med. 110: 145-152; 1987. Vollestad, N.K.; Wesche, J.; Sejersted, O.M. Gradual increase in leg oxygen uptake during repeated submaximal contractions in humans. J. Appl. Physiol. 68: 1150-l 156; 1990. Nishimura, D.G. Time-of-flight MR angiography. Mugn. Reson. Med. 14:194-201; 1990. Caputo, G.R.; Masui,T.;Gooding, G.A.W.;Chang, J.M.; Higgins, C.B. Popliteal and tibioperoneal arteries: Feasibility of two-dimensional time-of-flight MR angiography and phase velocity mapping. Radiology 182: 387-392; 1992. Caputo, G.R.; Kondo, C.; Masui, F.; Geraci, S. J.; Foster, E.; O’Sullivan, M.M.; Higgins, C.B. Right and left lung perfusion: In vitro and in vivo validation with oblique-angle, velocity-encoded Cine MR imaging. Radiology 180:693-698; 1991. Pelt, N.J.; Shimakawa, A.; Glover, G.H. Phase-contrast tine MRI. In: Book of abstracts: Eighth Annual Meeting of the Society of Magnetic Resonance in Medicine. Berkeley, CA:SMRM; 1989: 101. Walker, M.F.; Souza, S.P.; Dumoulin, C.L. Quantitative flow measurement in phase contrast MR angiography. J. Comp. Assist. Tomogr. 12:304-313; 1988.
10. Fisher, M.J.; Meyer, R.A; Adams, G.R.; Foley, J.M.; Potchen, E.J. Direct relationship between proton T, and exercise intensity in skeletal muscle MR images. fnvest. Radiol. 25:480-485; 1990. 1 I. Perktold, K.; Hilbert, D. Numerical simulation of pulsatile flow in a carotid bifurcation model. J. Biomed. Eng. 8:193-199; 1986. 12. Grant, E.G.; Tessler, R.N.; Perrella, R.R. Clinical doppler imaging. Am. J. Radio/. 152:707-717; 1989. 13. Greene, E.R. Noninvasive measurement of blood flow using pulsed Doppler ultrasound. In: J.A. Loeppky, M.L. Riedesel (Eds). Oxygen Transport to Human Tissues. New York: Elsevier/North Holland; 1982: pp. 135-150. 14. Frayne, R.; Rutt, B.K. Frequency response of Cine phase contrast. In: Book of abstracts: Eleventh Annual Meeting of the Society of Magnetic Resonance in Medicine. Berkeley, CA:SMRM; 1992: 2907. 15. Fronek, A.; Coel, M.; Bernstein, E.F. Quantitative ultrasonographic studies of lower extremity flow velocities in health and disease. Circulation 53:957-960; 1976. 16. Mackie, B.G.; Terjung, R.L. The influence of training on blood flow to the different fiber types in the rat skeletal muscle. Am. J. Physiol. 237:Cll l-Cl 18; 1983. 17. Petrofsky, J.S.; Phillips, C.A.; Sawka, M.N.; Hanpeter, D.; Stafford, D. Blood flow and metabolism during isometric contractions in cat skeletal muscle. J. Appl. Physioi. 50:493-502; 1981. 18. Guilfoyle, D.N.; Gibbs, P.; Ordidge, R. J.; Mansfield, P. Real-time flow measurements using echo-planar imaging. Magn. Reson. Med. 18:1-8; 1991. 19. Fleckenstein, J.L.; Haller, R.G.; Bertocci, L.A.; Parkey, R.W.; Peshock, R.M. Glycogenolysis, not perfusion, is the critical mediator of exercise-induced muscle modifications on MR images. Radiology 183:25-27; 1992.