Three-dimensional visualization of velocity profiles in the human main pulmonary artery with magnetic resonance phase-velocity mapping

Three-dimensional visualization of velocity profiles in the human main pulmonary artery with magnetic resonance phase-velocity mapping Detailed data on blood velocity fields in the normal human main pulmonary artery are an essential platform for discriminating physiologic from pathologic pulmonary flow patterns. Over the years, many studies have revealed quite inconsistent data mainly because of lack of suitable measuring techniques. By using combined cardiac- and respiratory-triggered magnetic resonance phase velocity mapping, very consistent data were obtained in 12 volunteers. In all subjects the location of the highest axial velocities was shifted from the inferior-right toward the superior-left part of the vessel area during the right ventricular contraction, with rapidly decreasing velocities to the inferior right evolving into retrograde flow in the deceleration phase. The mean temporal velocity profile was consistently skewed with a low flow region also toward the inferior-right vessel wall. The magnetic resonance phase shift method used in this study provided remarkably consistent high-quality data about human pulmonary artery velocity fields. This is most likely because of the use of combined cardiac and respiratory triggering. (AM HEART J 1994;128:1130-8.)

Erik Sloth, MD,a Kim C. Houlind, MSbv c Sten Oyre, MSb9 c W. Yong Kim, MD,b!c Erik M. Pedersen, MD, PhD,b, c Hans S. J$rgensen, MD,d and J. Michael Hasenkam, MD, DMScb~ c Aarhus, Denmark

Surprisingly little is known about the blood velocity profile in the human pulmonary artery (PA), although such data are essential for a basic understanding of the PA hemodynamics under physiologic and pathologic conditions. Furthermore, to obtain precise and reliable cardiac output (CO) measurements from pulsed Doppler ultrasound measurements in the pulmonary artery, detailed information about the velocity profile is necessary to estimate the temporal and spatial mean blood velocity. The sparse From the Departments of BAnaesthesia and bThoracic and Cardiovascular Surgery, %stitute of Experimental Clinical Research, and d Magnetic Resonance Center, Aarhus Kommune Hospital and Skejby Sygehus, Aarhus University Hospital, and the 8-dCardiovascular Research Center, Aarhus University. Supported by the Karen Elise Jensen Foundation, the Danish Heart Foundation, and the Laerdal Foundation for Acute Medicine. Received for publication Oct. 12, 1993; accepted Feb. 15, 1994. Reprint requests: Erik Sloth, MD, Department of Anaesthesia, Skejby Sygehus, Aarhus University Hospital, 8200 Aarhus N., Denmark. Copyright @ 1994 by Mosby-Year Book, Inc. 0002-8703/94/$3.00 + 0 4/l/58573

1130

information available on the flow pattern in the human PA is mainly the result of a lack of suitable methods for human studies. Different measurement techniques have been used in the past for both in vitro and in vivo experiments. These methods include laser Doppler anemometry (LDA), by which a blunt axial velocity profile has been found in an in vitro model of the PA,l, 2 and hot film anemometry (HFA) measurements which have been performed in human beings along the inferior-superior diameter.3 This study disclosed an almost flat mean temporal velocity profile with slightly higher velocities at the superior and inferior vessel wall compared with the centrally recorded velocities. Furthermore, HFA showed a flat velocity profile in the canine PA3; in pigs, the same technique revealed different results.4T 5 Intraluminal pulsed Doppler ultrasound (PDU) has been used widely in animal studies, which suggests the velocity profile in the canine PA to be skewed62 7 or of a more parabolic nature.8 Recently we found an almost flat mean temporal velocity profile in healthy

Volume 128, Number 6, ParI 1 American Heart Journal

Sloth et al.

1131

1. Two-dimensional modulus images of pulmonary artery. Left, Sagittal; right, transversal. Line used for double angulation intersecting mid pulmonary artery is shown.

Fig.

I. Hemodynamic parameters for volunteers

Table

retrog,,,l Volunteer no. 1

co (Llmin)

ALPHAS

TWnaJ TVLm

V?TUUl ~rne,,(,eak)

HR (min-I)

retwh,

ant,,,< f% i

antmax c%)

2924

3.9

14.1

1.4

1.7

6.0

18.5

1.6

1.6

40 70

0.6 0.2

2.1 0.5

1977

4322 4812 4009 4026

3.3 4.5 4.7

17.3 18.3 19.9

1.6 1.7

1.5 2.0

45 66

1.8 2.3

5.6 6.5

5

1158 1619

6 7 8

1023 2221 1340

3437 4478

4.5

20.7

1.5 1.6

1.5 1.6

63 74

0.4 1.0

1.6 3.9

1.4 1.1

65 60

0.4 1.1

1.9 6.7

1301

18.3 20.9 21.7

1.5 1.6

9

5.2 6.1 5.4

10 11 12

1885 1666

5.4 6.5

16.2 15.7

1.6 1.6 1.5

2.0 1.6 1.8

54 61

0.5 0.6 0.4

6.6 5.2 1.0

20.1 18.5 2.2

1.5 1.6

1.7 1.7

0.08

0.18

2 3 4

902 1738

3606 3831 4285 4585 4818

Mean

4094 550

SD

%mm,

46 75 60 11

1.2 0.9 0.6

2.6 2.6 2.6 6.1 3.6 2.0

mean Reynolds number; R.&.&, peak Reynolds number; ALPHA, Womersley’s o-parameter; CO, cardiac output; TVI,,ITVI,,, ratio imum to mean time velocity integral, VM,lV mean@pe,,k),peak systolic ratio of peak to mean spatial velocity; HR, heart rate, retrog,,/ant.y,, retrograde flow volume rate in percentage of antegrade systolic flow volume rate, retrog,,/ant,,,, maximum retrograde systolic Bow volume rate in percentage imum antegrade systolic flow volume rate. D/u, where V,,,, = spatial and temporal mean blood velocity in meters per second, D = internal pulmonary artery diameter in *hne* = VIO..” and Y = kinematic viscosity in square meters per second. D/u, where V, is the spatially averaged peak systolic velocity in meters per second. w%Jeak = V, SWomersley’s

a-parameter

= tan

x HR/GOU, where HR

I

of maxsystolic of maxmeters,

= heart rate (min-I).

adults with multiplane transesophageal Doppler echocardiography (MTEE) when measuring during end-expiratory apnea.g Magnetic resonance (MR) imaging is a noninvasive technique that offers the unique possibility for both tomographic imaging and accurate quantitative

blood velocity measurements in large vessels10-12and without any documented biologic side effects. Bogren et al. I3 used MR phase velocity (MRPV) mapping to study the velocity profile in healthy volunteers and in patients with pulmonary hypertension. They found a nonuniform flow pattern, especially in patients with

1132

Sloth et al.

American

December 1994 Heart Journal

Fig. 2. Typical example (volunteer 1) of three-dimensional visualization of temporal and spatial development of velocity profile viewed from downstream position. Top left, time in cardiac cycle is indicated on mean velocity curve with time interval relative to R wave in recorded electrocardiogram (milliseconds). Anatomic orientation shown in velocity profile a is used in successive plots. a, Beginning of cycle; b and c, acceleration phase; d and e, peak systole; f, g, and h, deceleration phase; i, diastole.

pulmonary hypertension. Later, Kondo et al.14 confirmed this in another MR study and demonstrated a not-flat velocity profile in normal subjects, further pronounced in patients with pulmonary hypertension. Based on these diverging reports in the literature, the aim of this study was to provide detailed and consistent information about the blood velocity profile in the human pulmonary artery in healthy, unsedated subjects using the MRPV encoding technique.

METHODS Population. Twelve healthy, unsedated young volunteers (23 to 33 years, mean 25.6 years) comprised the study group. The study was approved by the institutional committee on human research, and individual informed consent was obtained according to the Helsinki II declaration. Measurement technique and data aquisition. Arterial blood pressure and heart rate were measured before and after the MR flow measurements, and a deviation from the initial measurement of > 10 % in mean arterial blood pressure or heart rate led to exclusion of that particular series.

Volume 126, Number 6, Part 1 American Heart Joomal

Measurements were done on a 1.5 T 15s Gyroscan (Philips Medical Systems, Rest, Netherlands) with a spin-echo sequence for two-dimemhnal imaging and a flow adjusted gradient sequence for flow measurement.15 The MR investigations were performed with the volunteers in the supine position. To miniie respiratory synchronous PA movements and flow variations, all flow measurements were made during expiration by using selective respiratory gating. The respiratory signal was monitored on an oscilloscope to allow individually adjusted trigger gating. Cardiac triggering started data aquisition for the flow measurements 8 msec after the R wave in the electrocardiogram. The heart phase intervals depended on the heart rate and varied between 25 and 34 msec. From a transversal and sag&al modulus image (Fig. 1) a double-oblique plane of the mid PA was identified for velocity measurements, which were performed as two averages of 256 phase encoding steps. The velocity encoding was designed to measure the axial velocity components. The slice thickness was 8 mm and the pixel size was in the range of 1.4 mm2 to 5.5 mm2. The total measuring time was approximately 40 min for each volunteer, depending on both the heart rate and respiration frequency. Data analysis. The reconstructed and subtracted data were transferred from the MR host computer to a SPARC II workstation (Sun Microsystem, Inc., Mountain View, Calif.) for background phase error correction in a dedicated semiautomated software program.16Background noise were visually masked by assigning zero phase to pixels with a low-signal amplitude. This was followed by manual tracing of the PA modulus image to distinguish the PA flow from the surrounding cardiac structures. During this procedure, attempts were made to reduce PA movements in the coronal plane by visually determining a group of four center pixels in each traced area and assigning this group the same coordinates in the predefined matrix used for data storing. MR velocity data were transferred to a Macintosh computer (Apple Computer, Cupertino, Calif.) for graphic display. Using a standard spreadsheet program (Lotus l-2-3, Lotus Development Corporation, Middlesex, England), the velocities from each voxel in each phase were integrated, and the time velocity integral was multiplied with the cross-sectional area of the internal PA and the heart rate to compute the volume flow. From the MR velocity data the following hemodynamic parameters were calculated: mean Reynolds number (RE,,,), peak Reynolds number (RE,,,k) and Womersley’s a-parameter (Table I). To quantify the shape of the velocity profiles a spatial distribution index (V,&r,-(r&)) for the peak systolic velocity was calculated as the spatial maximum velocity (calculated as the maximum peak systolic value in the velocity matrix) divided by the spatial mean velocity (calculated as the spatial mean velocity during peak systole). To estimate the potential maximum error in single-point blood velocity-derived cardiac output (liters per minute) calculations resulting from a skewed velocity profile, the ratio of the maximum time velocity integral (TVI,,) to the

Sloth et al.

1133

3. Location in cross-sectional area where highest blood velocities in cardiac cycle were found for all 12 subjects. Fig.

mean time velocity integral was calculated (TV&J TVI,,). To quantitate the percentage of retrograde flow during systole, the index (retrogw$antw.J was calculated as the retrograde flow volume rate in percentage of the antegrade flow volume rate in systole (aystole defined as the first frame with net antegrade flow to the last frame before net retrograde flow occurred or to the frame before an increase in net antegrade flow if no net retrograde flow was seen). For the same time interval, an index retrog,,/antwaxwas calculated as the maximum retrograde flow volume rate in percentage of the maximum antepade flow volume rate. The MR velocity data were displayed as surface plots to provide a three-dimensional visualization of the velocity field at different times in the cardiac cycle (Fig. 2). A dynamic interpretation of the profile development was made possible by animation of consecutive velocity plots in rapid succession. Animation and graphic display was performed on a Macintosh computer with a commercially available software program (Spyglass Inc., Champagne, Ill.). The nomenclature used in this article refers to the anatomic orientation when the double-angulated modulus picture was viewed perpendicular to the main axis of the main PA. All plots were viewed from the downstream position. RESULTS

The gross hemoclynamic parameters for all volunteers including the mean and peak Reynold’s number, Womersley’s a-parameter, and cardiac output

Fig. 4. Images with superimposed velocity contours for each of 12 subjects during diastole when maximum retrograde flow area occurred. Separation areas are located toward inferior right and indicated by zero velocity contour. Positive and negative velocity contours are indicated for each subject. Anatomic orientation shown on image 1 is used on successive images. L, Left; R, right; S, superior; I, inferior.

are listed in Table I. An example of the three-dimensional visualization of the temporal and spatial development of the velocity profile from one volunteer representing the typical rotational pattern is seen in Fig. 2. During the beginning of the acceleration phase (a), the velocity profile was flat. Later in the acceleration phase (b, c) the highest velocities developed to the inferior right. Around peak systole Cd, e), the location of the highest velocities rotated clockwise (to the left) toward the inferior wall approximately 15 degrees. Early in the deceleration phase (f) the velocities decreased rapidly to the inferior right, continuing into retrograde flow later in the deceleration phase (g) and early diastole. This shifted the location of the highest velocities toward the superior/left vessel wall in the deceleration phase. In diastole the profile appeared flat.

In three subjects, the highest velocities occurred to the inferior right without any obvious rotation. In two measurement series the highest velocities developed more to the inferior, also without obvious rotation. The clockwise rotation seen in seven subjects was from approximately 15 to 60 degrees. In one subject clockwise rotation of approximately 30 degrees was followed by a clearly counterclockwise rotation of approximately 40 degrees. The peak antegrade velocities were predominantly located close to the inferior vessel wall. (Fig. 3). The shift of the highest velocities from the inferior-right part toward the superior-left part of the vessel area was observed in all 12 volunteers. The retrograde flow pattern was also a consistent finding in all measurement series (Fig. 4). Retrograde flow started as early as 53 msec (range 53 to 310 msec, median 135 msec) after peak

volunw 11, Hwmbaf6, Part 1 AtnmicanHaaclJournal

Sloth et al.

1135

5. Three-dimensional visualization of mean temporal velocity profile shows low velocity area toward inferior right. Profile is from same measurement series shown in Fig. 2.

Fig.

systole and continued throughout early diastole, being most pronounced at the inferior-right vessel wall (Fig. 4). The percentage of retrograde flow is seen in Table I. The late diastolic phase was characterized by mostly antegrade flow without any consistent pattern. In all volunteers the mean temporal velocity profile was consistently skewed with a well-defined low velocity region toward the inferior right (Figs. 5 and 6). A quantitative description of the deviation from a non-flat mean temporal velocity profile revealed a ratio of the maximum time velocity integral to the mean time velocity integral of 1.6 + 0.08 (mean + SD; Table I). The peak systolic index calculated as the ratio of the maximum/mean velocity indicated a non-flat, cross-sectional velocity distribution because the values were different from unity (Table I). DISCUSSION

In healthy subjects the venous return, and hence the right ventricular preload, changes during the respiratory cycle. This is further augmented during mechanical ventilation, which is often used in animal studies in which velocity profiles have been studied. Compromising the venous return to the right atrium might explain to some extent the diversity in PA flow

patterns described in the literature. Lucas et al.6 showed in a canine model that experimentally induced atrial septum defects influenced the flow pattern in terms of altering the location for the maximum recorded velocity. Pulmonary hypertension, caused or followed by increased blood flow, has been shown both in humanla 14,17 and animal studie.& 7* “7 is to imply profound changes in the flow pattern. Consequently, respiratory standardization should be used when detailed velocity profile evaluation in the pulmonary vascular system is intended. Lack of such respiratory standardization may be the major reason for the nonuniformity in normal human flow pattern described by Bogren et alI3 Kondo et a1.14clearly demonstrated differences in flow pattern between subjects with and without pulmonary hypertension. From our own pilot studies in which we performed measurements in one volunteer with and without respiratory triggering, it was obvious that the measurement series without respiratory triggering showed blurring of the velocity patterns and contained much more noise than the respiratory-triggered series. This is not unexpected because the sampling scheme of the MR sequences anticipates stable hemodynamics throughout the aquisition time and as in-plane motion induces artifacts. Also, the

1136

Sloth et al.

American

December 1994 Heatl Journal

c

4

6. Mean temporal velocity images from all 12 subjects. Inferior-right colors appear lighter, indicating low velocity area. Anatomic orientation shown on Image 1 is used in successive images. I, Inferior; S, superior; R, right; L, left.

Fig.

overall characteristics of the untriggered series differed significantly from the general findings in all other measurement series including its own control by using respiratory triggering. A main problem caused by the respiratory gating is increased measuring time, which has to be remembered if this method is applied to a clinical set-up. Along with a high spatial resolution used in this study, MR phase velocity encoding is unique in providing detailed velocity information across the entire vessel area. These features make this method preferable to other methods used in vivo so far, especially in the PA, a curved vessel in which complicated three-dimensional velocity patterns are to be expected.20 In the future, application of pulse sequences with shorter echo times, like the free induction decay-acquired ethos (FACE) method, could further enhance the accuracy of these detailed velocity measurements.21 The impact of PA movements in the coronal plane was minimized by giving the center of each traced area the same coordinates in the storing matrix. Flow variations from axial translation could not, however, be avoided in this study. According to the findings of S#mod et al.,22 who found no significant changes in the velocity profile from one to two diameters downstream of the porcine pulmonary annulus, axial

translation is considered less important. Bogren et all3 described a shift in the flow axis from the mid to the distal part of the PA. These flow variations might be induced by the bifurcation as described from in vitro studies.rp 2 Bogren et all3 reported a nonuniform rotation of the velocity profile during the right ventricular ejection phase in healthy human beings. Although the rotational feature was the least consistent characteristic in the present study, 7 of 12 measurement series showed a clockwise rotation in the acceleration phase. The cause of this rotation is not quite clear but may partly be attributed to the anatomy and contraction pattern of the right ventricle and outflow tract because the curvature itself should give rise to two counter-rotating vortexes.20 Whether a higher temporal resolution might have disclosed slightly further rotational patterns is not clear. The shift of the highest velocities from the inferior-right toward the superior-left part of the vessel area and the appearance of retrograde flow to the inferior right during the right ventricular ejection is in agreement with anima16~ 1a,22 and human studies with comparable pulmonary artery characteristics97 13,141ls Retrograde flow early in systole has been described in patients with pulmonary hypertension and has been

*

volume 126, Number 6, Part 1 American Heart JOumal

Sloth et al. 1137

400

o Net flow 7

300

v Retrograde flow Antegrade flow

l

f X

E 200 a, s 3 g

100

5 z 0

-100

0

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

2

4

6

8

10

12

14

16

18

20

22

24

26

28

30

32

Frame number Fig. 7. Typical volume flow rate curve (volunteer 1). Nef flow, Retrograde flow + Antegrade flow.

shown to occur later in systole in healthy subjects.13, ‘* We found retrograde flow early after peak systole, probably the result of the use of respiratory triggering, because this is the only major difference in experimental set-up compared to other MR studies.13>‘* These events are in agreement with theoretic considerations for flow development in a curved vessel when a flat inlet profile is assumed.20 This flow pattern indicates a late systolic vortex formation that exhibits retrograde flow at the inner curvature of the vessel. Retrog,&m&r,, and only accounted for 0.9 % + 0.6% retrog,ax/ant,ax and 3.6% rfr 2.0% (mean f SD; Table l), respectively. Retrograde flow late in systole and early diastole may preserve closing of the pulmonary valves as described in aortic valve closure.23 This may explain the net retrograde flow seen in 10 of 12 subjects in the current study (Fig. 7). Whether the retrograde flow proceeded across the pulmonic valves cannot be evaluated from the present study because only measured in the mid PA location. Valid conclusions of diastolic flow patterns can not be drawn from this study or from other studies found in the literature. The finding of a consistent mean temporal velocity profile in all subjects is important in relation to cardiac output measurement based on pulsed Doppler blood velocity recordings in the human PA. Such

measurements have lately been given increasing attention in an attempt to develop a reliable noninvasive method for cardiac output measuremet.24‘27 The skewed mean velocity profile demonstrated in this study may induce erroneous cardiac output measurements when using Doppler echocardiography to estimate the mean velocity based on single-point recording. The ratio TVI,JI’VI,,, in Table I, which is in the range 1.4 to 1.7, expresses the maximum error in estimating cardiac output from a single-sample volume corresponding to a square of four pixels. This is in contrast to our own findings with MTEE, where this error varied from 1.1 to 1.4.g For exact comparison of these two sets of data, the analyzing method (gi&ng the center of each PA tracing the same coordinates in the data storing matrix) should not be used for the MR data because it cannot be applied to the single-point Doppler data. Future studies should include the measurement of in-plane velocities to evaluate all three velocity components to fully understand the three-dimensional nature of the flow in the PA. Conclusion. Evaluation of the pulmonary artery velocity fields in nonanesthetized human beings with combined respiratoryand cardiac-triggered MR phase velocity encoding has produced detailed and consistent results. We found (1) a consistent shift of

1138

December 1994 American Heart Journal

Sloth et al.

the highest velocities from the inferior-right toward the superior-left part of the human pulmonary artery cross-sectional vessel area during the right ventricular ejection; (2) rapidly decreasing velocities toward the inferior right evolving into retrograde flow in the deceleration phase; (3) a consistently skewed mean temporal velocity profile with a well-defined low flow region also located to the inferior right, and (4) basic hemodynamic parameters in the normal human PA are given, The present data provide a valuable basis for clinical studies comprising cardiac output measurement and pathologic pulmonary artery flow patterns.

REFERENCES

1. Sung HW, Yoganathan AP. Axial flow velocity patterns in a normal human pulmonary artery model: pulsatile in vitro studies. J Biomech 1990;23:201-14. 2. Ganiin JM, Sung HW, Yoganathan AP, Ball J, McMillan S, Henry WL. Doppler flow velocity mapping in an in vitro model of the normal pulmonary artery. J Am Coll Cardiol1988; 12:1366-76. 3. Reuben SR, Swadlmg JP, Lee G de J. Velocity profiles in the main pulmonary artery of dogs and man, measured with a thin-film resistance anemometer. Circ Res 1970;27:995-1001. 4. Paulsen PK. The hot-film anemometer-a method for blood velocity determination. II. In vivo comparison with the electromagnetic blood flowmeter. Eur Surg Res 1980;12:149-58. 5. S#mod L, Hasenkam JM, Kim WY, Nygaard H, Paulsen PK. Tbree-dimensional visualisation of velocity profiles in the normal porcine pulmonary trunk. Cardiovasc Res 1993; 27:291-5. 6. Lucas CL, Henry GW, Ferreiro JI, Ha B, Keagy BA, Wilcox BR. Pulmonary blood velocity profile variability in open-chest dogs: iniluence of acutely altered hemodynemic states on profiles, and influence of profiles on the accuracy of techniques for cardiac output determination. Heart Vessels 1988;4:65-78. 7. Frants EG, Henry GW, Lucas CL, Keagy BA, Lores ME, Criado E, Ferreiro JI, Wilcox BR. Characteristics of blood flow velocity in the hypertensive canine pulmonary artery. Ultrasound Med Biol1986,12:379-85. 8. Henry GW, Johnson TA, Ferreim JI, Hsiao HS, Lucas CL, Keagy BA, Lores ME, Wilcox BR. Velocity profile in the main pulmonary artery in a canine model. Cardiovasc Res 19&1; 18~620-5. 9. Sloth E, Pedersen EM, Nygaard H, Hasenkam JM, Juhl B. Multiplane transasophageal Doppler-achocardiograpraphic measurements of the velocity profile in the human pulmonary artery. J Am Sot Echocardiogr 1994;7:132-40. 10. Underwood SR, Firmin DN, Klipstem RH, Raas RS, Longmore DB. Magnetic resonance velocity mapping: clinical application of a new technique. Br Heart J 1987;57:404-12.

11. Firmin

DN, Nayler

GL, Klipstein RH, Underwood SR, Rees RS, Longof MR velocity imaging. J Comput Assist Tomogr 1987;11:751-6. 12. Van Rossum AC, Sprenger M, Visser FC, Peels KH, Valk J, Roes JP. An in vivo validation of quantitative blood flow imaging in arteries and veins using magnetic resonance phase-shift techniques. Eur Heart J

more DB. In vivo validation

1991;12:117-26.

13. Bogren HG, Klipstein RH, Mohiaddin RH, Firmin DN, Underwood SR, Rees RS, Longmore DB. Pulmonary artery distensibility and blood flow patterns: a magnetic resonance study of normal subjects and of patients with pulmonary arterial hypertension. AM HEART J 1989;118:990-9. 14. Kondo C, Caputo GR, Masui T, Foster E, O’Sullivan M, Stulbarg MS, Golden J, Catterjee K, Higgins CB. Pulmonary hypertension: pulmonary flow quantitication and flow profile analysis with velocity-encoded tine MR imaging. Radiology 1992;183:751-8. 15. Green JP, van Dijk P, In den Kleef JTE. Design of flow adjustable gradient waveforms [Abstract]. Proceedings of the 6th Annual Meeting of the Society of Magnetic Resonance in Medicine, New York. 19&37$68. 16. Walker PG, Cranney GB, Scheidegger MB, Waseleski G, Pohost GM, Yoganathan AP. Semiautomated method for noise reduction and background phase error correction on MR phase velocity data. J Magn Re son Imaging 1993;3:521-30. 17. Woletanholme GEW, Knight J. Circulatory and respiratory mass transport. London: Churchill, 1969172-99. 18. Katayama H, Henry GW, Lucas CL, Ha B, Ferreiro JI, Frantz EG, Kneski R. Three-dimensional visualization of pulmonary blood flow velocity profiles in lambs. Jpn Heart J 1992;33:95-111. 19. Okamoto M, Miyatake K, Kinoshita N, Sakakibara H, Nimura Y. Analysis of blood flow in pulmonary hypertension with the pulsadDoppler flowmeter combined with cross sectional echocardiography. Br Heart J 19&1;51:407-15. 20. Lanser P, Yoganathan AP. Vascular imaging by color Doppler and magnetic resonance imaging. Berlin: Springer-Verlag, 1991:62. 21. Scheidegger MB, Maier SE, Boesiger P. FID-acquired ethos (FACE): a short echo time imaging method for flow artefact suppression. Magn Reson Imaging 1991;9:517-24. 22. S$mod L, Pedersen EM, Kim WY, Hasenkam JM, Nygaard H, Paulsen PK. Axial development of velocity fields in the porcine main pulmonary artery system. Heart Vessels 1994;9:67-78. 23. Hatle L, Angelsen B. Doppler ultrasound in cardiology. Physical principles and clinical applications. 2nd ed. Trondheim: Las & Febiger, 19851-331. 24. Muhiudeen IA, Kuecherer HF, Lee E, Cahalan MK, Schiller NB. Intraoperative estimation of cardiac output by transesophageal pulsed Doppler echocardiography. Anesthesiology 1991;74:9-14. 25. Savino JS, Troianos CA, Aukburg S, Weiss R, Raichek N. Measurement of pulmonary blood flow with transesophageal two-dimensional and Doppler echocardiography. Anesthesiology 1991;75:445-51. 26. Roewer N, Bednarz F, Schulk AM, Esch J. Continuous measurement of intracardiac and pulmonary blood flow velocities with transesophageal pulsed Doppler echocardiography: technique and initial clinical experience. J Cardiothorac Anesth 1987;1:418-28. 27. Gorcsan J, Diana P, Ball BA, Hattler BG. Intraoperative determination of cardiac output by transesophageal continuous wave Doppler. AM HEART J 1992;123:171-6.