Evaluation of Mitral Regurgitation by Cine Magnetic Resonance Imaging Gerard Aurigemma, MD, Nathaniel Reichek, MD, Mark Schiebler, MD, and Leon Axel, PhD, MD, with the technical assistance of Christine Harris, RT, and Ali Muhammad
We used tine magnetic resonance imaging (MRI) to assess mitral regurgitation (MR) in 40 patients with coronary and/or valvular disease and 10 normal subjects and compared results to pulrsed (n = 30) or color flow Doppler mapping (n = 20). Mitral regurgitation produced a dynamic signal void in the left atrium in systole in 15 of 16 patients with MR by pulsed Doppler and in an additional 15 of 16 patients whose MR was demonstrated by color flow Doppler. There were no false positives (sensitivity 54%, specificity 100% for both). The ratio of single-plane, maximal jet area to left atrial area was used to grade MR severity with mild defined as 40%. Cine MRI classification was identical to pulsed Doppler echocardiography in 26 of 30 patients and to color flow Doppler in 16 of 20 patients with no differences of >1 grade. Cine MRI consistently depicted smaller flow disturbances than pulsed Doppler (slope = 0.65) or color flow Doppler (slope = 0.60). Nonetheless, the tine MRI area ratio correlated well with pulsed Doppler (r = 0.78) and with color flow Doppler (r = 0.74). Thus, planar analysis of tine MRI in patients with MR of varying severity gave results that were similar to Doppler echocardiography. At present, for routine clinical assessment of MR, the benefits of tine MRI may be limited to patients in whom transthoracic Doppler echocardiography is not adequate. (AmJ Cardiol 1990;66:621-625)
From the Cardiovascular Section, Department of Medicine and the Devon Imaging Center, the Department of Radiology. Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania. This study was supported in part by the Commonwealth of Pennsylvania Health Services Contract and by a Fellowship Grant from the Southeastern Pennsylvania Chapter of the American Heart Association, Philadelphia, Pennsylvania. Dr. Aurigemma is now at the Department of Medicine, Division of Cardiology, University of Massachusetts Medical Center, Worcester, Massachusetts; Dr. Schiebler is now at the Department of Radiology, University of North Carolina Medical Center. Chapel Hill, North Carolina. Manuscript received April 10, 1990; revised manuscript received and accepted April 30, 1990. Address for reprints: Gerard Aurigemma. MD, Division of Cardiology. University of Massachusetts Medical Center, 55 Lake Avenue North, Worcester. Massachusetts 01655.
ine magnetic resonance imaging (MRI) uses gradient refocused echoes to produce dynamic, high-resolution tomographic images of cardiac motion and blood flow without contrast administration. I The heart can be imaged at multiple levels simultaneously and the resulting images displayed in tine loop fashion. Moreover, turbulent flow associatedwith valvular regurgitation appearsas a discrete signal void2,3and is readily distinguishable from laminar flow. Pulsed and color flow Doppler echocardiography are used clinically for the evaluation of mitral regurgitation (MR), but can sample only selectedplanes through the heart4 Because tine MRI can tomographically sample the entire cardiac volume and distinguish turbulent from laminar flow, it may be valuable to apply this technique to the study of MR. Our study assessedthe ability of tine MRI to detect MR and to quantitate the size of the associated flow disturbance as compared to pulsed and color flow Doppler.
C
METHODS Patients:
Forty patients aged 25 to 88 years with a medical history of valvular or coronary artery disease and 10 volunteers aged 25 to 70 years without known heart diseasemade up the study group. Patients with known valvular heart diseasewere approached for participation in the study based on results of a recent Doppler echocardiographic examination or at the suggestion of their physician. In addition, patients with clinically normal valves who had a tine MRI examination for evaluation of coronary artery bypass graft patency5 were also included in the study group. All subjects were clinically stable, in regular sinus rhythm and gave informed consent. The study was approved by the institutional review board of the Hospital of the University of Pennsylvania. Cine magnetic resonance imaging technique: MRI was performed in a 1.5 Tesla Signa (General Electric) MRI system. Each study began with electrocardiographically-gated multislice coronal spin echo images of the thorax for cardiac localization. The image matrix was 128 X 256, interpolated to 256 X 256 for display and the field of view was 32 or 40 cm depending on patient size. The coronal serieswas used to define the extent of the left atrium. At least 2 tine MRI seriesof two to four 5 to 10 mm-thick axial slicesthrough the left atrium and left ventricle with interslice distances of 5 mm to 2 cm were obtained in all subjects.Early in the series,however, we concluded that the optimal imaging approach consisted of interleaving 3 sets of three 10
THE AMERICAN
JOURNAL
OF CARDIOLOGY
SEPTEMBER
1, 1990
621
mm-thick slices with an interslice distance of 20 mm per series; this approach was used routinely for the patients in the remainder of the series. The tine MRI techniques used flip angles of 30’ and gradient refocused echoes with an echo time of 12.5 ms and repetition time of 22 ms. The radio frequency pulse, readout gradient, phase-encoding gradient and slice-selective gradient occurred independently of the cardiac cycle every 22 ms. The R wave of each electrocardiographic cycle controlled advancement of the phase-encoding gradient to the next image line. In this study, frame rate was determined by dividing the RR interval by the repetition time and then by the number of slices desired in the tine loop. Thus, for an RR interval of 800 ms (heart rate 75 beats/mm) 36 frames were available. If 3 slices were desired, each tine loop contained 12 frames per cardiac cycle. The number of frames per cardiac cycle for subjects in this series ranged from 5 to 26. It is also possible to obtain more frames by interpolation in the image reconstruction. Two to 4 signal averages per image series were obtained to optimize image quality. Each acquisition took 128 cardiac cycles, 1 for each Y line in the image. Time per acquisition at the time of this study included 2 to 3 minutes for prescan, 3 to 5 minutes for imaging and 5 to 10 minutes for imaging processing. Total time per study was approximately 50 minutes. Subsequent software improvements permit patient time in the system of roughly 35 minutes with image reconstruction time of 3 minutes per series.
Doppler techniques: Thirty subjects were evaluated using standard 2-dimensionally guided pulsed Doppler mapping of MR using commercially available systems. A 2.5 MHz Doppler flow probe was used to perform examinations from apical and parasternal windows. A point-by-point meticulous search for disturbed flow was conducted from apical 2- and 4-chamber and parasternal long-axis views in all patients. Twenty subjects were evaluated using color flow Doppler mapping with a Hewlett-Packard model 77020 system. A careful search was made in each view to record the maximum systolic left atria1 area of disturbed flow. Evaluatiin of rewlts--eine imaging: Qualitative recognition
magnetic
resonance
of MR by tine MRI was arrived at by consensus at a reading session attended by 2 cardiologists and 2 radiologists. Jets of MR were defined as dynamic zones of signal void in the left atrium in systole extending from the mitral valve plane. For each study, the tine MRI frame showing the maximal signal void associated with MR was selected. Quantitation was performed by digitizing the area of the signal void and the left atria1 area from film hard copy using a microcomputer equipped with a backlit, transparent high-resolution digitizer (Hewlett-Packard 9825A). Because the imaging planes were different by tine MRI and Doppler echocardiography, the ratio of maximal jet area to left atria1 area was used to normalize results. Severity was graded as follows: mild, jet area to left atria1 area ratio lgl-ade.
622
THE AMERICAN
JOURNAL
as an area ratio as HO%. Classifi-
OF CARDIOLOGY
VOLUME
66
isunifOrmhighsignalthfw&fd all 4 cardiac chambers and etructurefi have attemmtin the denceding aorta. Myocardial ed signal compared to bbod pool. Moderator band is visualiz0d in right venbide. B, arrow points to dosed mitral valve. C and D, in later systole, uniform high signal persists in lett atrium and systolic s~pt~l thickening is also shown. A = descendingaorb;LA=leftatriun;LV=leRventride;RA=ri%lt atriun;RV=rightventdcle.
area to left atria1 area greater than 40%, after the work RESULTS of Helmcke et al.” Mitral regurgitation was detected by Doppler echoDoppler echocardiography: Qualitative recognition cardiography in 32 of 50 subjects and by tine MRI in of MR by Doppler echocardiography was determined 30 of 50 subjects. Mitral regurgitation was found in 16 by 2 observers who were unaware of the tine MRI re- of the 20 subjects examined by color flow Doppler and sults. A single observer, blinded to the tine MRI results, 16 of 30 examined by pulsed Doppler (Figure 1). Cine assessed the severity of MR. Mitral regurgitation was MRI detected MR in 15 of 16 subjects with MR by defined as a broad high-velocity spectrum with aliasing pulsed Doppler and 15 of 16 with MR by color flow detected in the left atrium and extending from the valve Doppler (Figure 1). There were no false positives. Overplane in systole for pulsed Doppler. Using color flow all sensitivity was 94% and specificity was 100% with Doppler, MR was defined as a retrograde flow region Doppler flow mapping as a reference standard. with or without variance and aliasing extending from Using tine MRI, intracardiac laminar blood flow the mitral valve into the left atrium in systole and con- generates a higher signal than myocardial structures firmed by pulsed Doppler. The area of the MR color (Figure 2). A regurgitant jet of MR was visualized as a flow disturbance was outlined on clear plastic film and signal void in the left atrium in systole (Figure 3). The digitized. In subjects who had pulsed mapping alone, a extent, shape and direction of the signal void varied perimeter of the zone in the left atrium where turbulent from patient to patient. Also, there often was dynamic flow was detected was made on a clear plastic overlay variation in the signal void size and shape in a single and digitized. The jet area to left atria1 area was ex- plane throughout systole (Figure 4). pressed as a ratio. The severity of MR was graded as Classification: Cine MRI classification was identical mild if the ratio was 40%.” tical to color flow Doppler in 16 of 20 (80%) subjects. There were no differences of >l grade. Statistical analysis: For the purposes of this study, Doppler echocardiography was taken as the reference Jet area/left atrial area comparison: Figure 5 shows standard. Sensitivity and specificity of tine MRI com- % left atrium occupied by the MR jet by color flow pared to color flow Doppler and pulsed Doppler were Doppler and tine MRI. The regression equation was calculated. The classification of MR as absent, mild, tine MRI = 3.8 + 0.60 color Doppler (r = 0.74). For moderate or severe by tine MRI and pulsed and color pulsed Doppler, the relation was tine MRI = 4.6 + flow Doppler on single-plane maximum jet area left 0.65 pulsed Doppler (r = 0.78) and is shown in Figure atria1 area ratio was compared to define semiqualitative 6. Cine MRI, using the technique described, therefore agreement. In addition, absolute jet area/left atria1 cor- consistently depicted a smaller relative area of flow disrelations were compared by linear regression.
FlGURE 3. Axial tine magnetic remname images taken from the study of a patient with moderate mitral regurgitation. Grientation is similar to Figure 2 for these systolic images. A, /afge afrow paints to wedge-shaped signal void in left atrium dire-&d post&&y. Small arrows point to closed mitral valve leaflets. B, C, D, shape of signal void does not appear to change significantly in these systolic images.
FIGURE 4. Axial systolii tine magnetic resonance images from study of a patient with severe mitral regurgitath. A, /afge whife arrow points to signal void in left atrium emanating from region of dosed mitral kafkts (sma// arrows). Et, C, D, area of signal void enlarges and propagates medially and postthroughout systolo. In C and D, signal void has diminished in intensity but appears to occupy a larger perceh age of the left atria1 area, including pulmonary veins.
THE AMERICAN
JOURNAL
OF CARDIOLOGY
SEPTEMBER
1, 1990
623
turbance when compared to pulsed or color flow Doppler mapping. DISCUSSION
True volumetric estimation of the severity of valvular regurgitation can only be accomplished by comparing forward left ventricular stroke volume with total left ventricular stroke volume. Such estimates of regurgitant volume are only meaningful in patients with single-valve regurgitation. While the clinically accepted method for computing regurgitant volume subtracts Fick (or thermcdilution) cardiac output from angiographic cardiac output,6 regurgitant volume has also been estimated noninvasively by radionuclide angiography7 and by tine computed tomography.8 In addition, 2-dimensional9 and Doppler echocardiographytO have been used to compare right and left ventricular stroke volumes and to estimate regurgitant fraction. However, for MR, clinicians commonly rely on semiquantitative assessment of the area of flow disturbance, such as is provided by Doppler echocardiography or by angiographic (1 +, 2+) grading. In general, there is a good correlation between Dopplerderived regurgitant jet area and qualitative angiographic assessment of regurgitation.4 Because tine MRI, like color flow Doppler mapping, displays disturbed cardiac flow, it appears to be a suitable method for the semiquantitative grading of MR. Cine magnetic
resonance
imaging
signal
properties:
With gradient refocused imaging, myocardial structures 50
cMR1= 3.6 t 0.60 CFD r=.74 40
have attenuated signal relative to blood because their less mobile protons remain in the imaging plane and are continuously excited. The signal intensity of flowing blood depends, in part, on its velocity and on whether the flow is laminar or turbulent. In general, laminar flowing blood generates uniform, bright signal; this may be due to the continuous entry into the imaging plane of fresh blood whose spins are “unsaturated” and give maximal signal.tJ’ Factors that may also contribute to the high signal are the high-proton density of blood2 and the lack of phase cancellation induced signal loss and short echo times. The regurgitant jet of MR was associated with loss of signal in a region of the left atrium in systole. Multiple mechanisms of signal reduction may exist. Dephasing of spins caused by shear between flow layers moving at different velocities adjacent to heart or blood vessel wall can produce a signal void. Time of flight effects are thought to be minimal for tine MRI. Fram et all2 have shown the velocity dependence of signal amplitude at short repetition times in an in vitro model with evidence of a decrease in signal at higher velocities. For regurgitant jets, signal voids are presumably due to turbulence13; phase cancellation of signal may occur among protons moving at different velocities.2 Cine magnetic resonance imaging/Doppler comparisons: Overall qualitative correlation between tine MRI
and Doppler echocardiography was good with no differences of > 1 grade. The quantitative correlation was also good, though tine MRI appeared to show smaller flow disturbances that pulsed Doppler (slope = 0.65) or color flow mapping (slope = 0.60). This discrepancy may be due to multiple factors. First, temporal resolution of
.
. .
60 .
cMR1= 4.6 t 0.65 PO i-s.78
50
10
0
0
I
10
.
I . I 20 30
I ’ 40 50
0 3
PULSED DOPPLER
COLORFLOWDOPPLER FIGURE 5. Comparison between jet areahefi atrlal area ratio (%LA) derived from cdor Raw Doppler (CFD, x-axis) ad from mmaance Imaging (cMRI, y-axis). cMRl = tine einemagnetk-luaimaging.
624
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
66
FIGURE
6. Comparison
between
jel area/left
atrial
area ratio
derived lmm pulsed Doppler (x-axis) and from tine magnetk resonance imaging nance imaging.
(cMRI,
y-axis).cMRI
= tine magnetic
reso-
mapping was different for tine MRI than for pulsed Doppler. Thus, the maximal jet area defined for pulsed Doppler was mapped throughout systole and the composite of all points positive for MR were taken. Similarly, for color flow Doppler, the frame rates ranged from 15 to 30/second. In contrast, as noted before, tine MRI has an effective frame rate of only 12/cardiac cycle/ slice in a 3-slice acquisition when the heart rate is 75 beats/min. Because MR flow varies over time, the differing temporal resolution of sampling by Doppler echocardiography could result in larger apparent jet area even if both Doppler echocardiography and tine MRI jets were actually identical. Secondly, differences in imaging planes used could affect the results, because axial tine MRI images lie closer to the short-axis plane of the MR jet, while the apical Doppler window lies closer to the jet’s long-axis plane. Recent software enhancements permit oblique imaging with tine MRI and further studies should permit comparison of comparable image planes. Additionally, while Doppler and tine MRI both detect turbulent flow, they may in fact be measuring different physical phenomena. Finally, discrepancies between tine MRI and Doppler grading of MR may have been due to differences in loading conditions between the time of the 2 studies. Patients had their tine MRI and Doppler studies on the same day whenever possible. Nonetheless, because hemodynamic monitoring was not performed during MRI, hemodynamic changes, with attendant effects on regurgitant jet area, may have taken place. Study limitations: There are a number of important limitations to this study. The size of the color flow Doppler jet of MR has been shown to vary depending on technical factors, including gain adjustment and filter settings.4,14Technical factors also influence the size of the flow disturbance as depicted by tine MRI; the size of the signal void associatedwith valvular regurgitation appears to be dependent on echo time. In addition, as noted before, tine MRI frame rate is dependent on the patient’s heart rate and number of slices per acquisition. Finally, window and threshold settings for the tine MRI display were set to provide maximum contrast for each image; no attempt was made to standardize these settings. This study may also be criticized for the use of Doppler echocardiography as a reference standard. The limitations of Doppler semiquantitative grading have been well described.15However, we chose to compare the results of tine MRI grading to Doppler echocardi-
ography becausethe latter remains the standard noninvasive method used to evaluate valvular regurgitation. Acknowledgment: We wish to thank Kathleen Bogin, RN, and Margaret Dunlop, RN, for their patient recruitment and data management, Robert Morrow for his assistancein processingthe Doppler data, the technical staff of the Devon Center for their assistance,and Janine Carter and Beverly Butler for their help in the preparation of the manuscript.
REFERENCES 1. Sechtem U, Pllugfelder PW. White RD, Gould R, Halt W, Lipton M, Higgins C. Cine MR Imaging: potential for the evaluation of cardiovascular function. Am J Kudiology 1987;I 48:239-246. 2. Schiebler M, Axcl L, Reichek N, Aurigemma G, Yeager B, Douglas P. Bogin K. Kreaael H Correlation of&x MR imaging with 2-dimensional pulsed Doppler echocardiography in valvular inaufliciency. J Compur .4ssisr Tomogr /987;/ I627-632. 3. Sechtem U, Pllugfelder PW, Cassidy MM. White R, Cheitlin M, Schiller N, Higgins C. Mitral or nortic regurgitatmn: quantification of regurgiwnt volumes with tine MR imaging. Radrology /988:/67.425-430. 4. Helmcke F. Nanda NC, Hsiung MC, Soto B, Ade) C. Goyal R, Gatewood R. Color Doppler aaaessment of mitral regurgitation with orthogonal planes. Circulalion l987;75.175~lU3. 5. Aurigemma G, Relchek N, Axcl L, Schiebler M, Ilarris C, Kressel H. Nomn\aswedeterminatlon ofcoronary artery bypass graft patency bq tine MRI. Circulution 1989,80:1595-1602. 6. Croft CH, Lipscomb K, Mathia K. Firth B, Nlcod P. Tilton G, Winniford M, Hillis L. Lmitations of qualitative angiographx grading in aortic or rmtral regurgitation Am J Cardiol 1984;S3:1593-1598. 7. Kelback H. Alderahvile J, Svcndscn JH, Folke K, Nielsen SL, Munck 0. Combined first pars and equilibrium radionuchde cardiographic determination of stroke volume for quantitation of valvular regurgitation J Am Co// Curd;01 I9XX:I I ,769 773 8. Rater S, Rumbcrgcr .I. Feiring A. Stanford W, Marcus M Prccislon measuremen& or right and left ventricular volume by tine CT. Circulution /986:74:890900. 9. Biumlein S, Bouchxd A, Schdler NB, Dac %I, Bird B. Ports T , Botvimk E. Quantltation of mitral regurgitation by Doppler cchocardlography. Circu/a~io,r 1986:74.306 314. 10. Rokcy R, Sterling K. Logbhi W. Sartorl M. L.lmacher M. Kuo L. Quinones M. Determination of regurgitant fraction m wlated mitral oraortic regurgitation by pulsed Doppler two-dm~cn&~nal cchocardlography. J .4n1 (bll C’ardrol /9X6,7.1?73 127x. 11. Aurlgemma GP, Reichek N, AxeI I., Kreascl H. Cudiac tine MRI: normal Intracardiac and great VCS~CI Ilox. C/in KG IY87.35.25YA 12. Fran E. Hedlund I.. Dlrnlck R, Glover G, Herlkens R. Parameters dctermining the signal of flowing fluid in gradient refocused imagmg: flow velocity, TR. and nip angle. Prow~d~n~s of thr Soort~~fur Ma~nrri~~ Krsotronce vz Mrdrcinp Vth Anma/ Mrctin~, /Y&5. 1.84. 13. Evans A. Blmder R, Her&ens R. Sprit/cr C, Kuethe D, Fran1 E, Hedlund L. F.ffccta of turbulence on signal intenslt) in gradxnt echo mugcs. Iwe~t Radio/ lYXX.23.517 51x. 14. Wang M, Matsut~~ura M. Surukl h, Omoto R. Tcchnicul and biologic wurcc~ ofvariability III the mapping of axtIc. mitral and tricuspid color now jets. .4nt J Cbrdrol 1987;60.-847 851 15. Bolgcr AF, Eiglcr %L. hlaurer G Qudllt~f)~llg\;il\ulllr regurgltatmn: limltattons and inherent ;wumptwnruf Doppler tcchniqucr. (;r
We used tine magnetic resonance imaging (MRI) to assess mitral regurgitation (MR) in 40 patients with coronary and/or valvular disease and 10 normal subjects and compared results to pulrsed (n = 30) or color flow Doppler mapping (n = 20). Mitral regurgitation produced a dynamic signal void in the left atrium in systole in 15 of 16 patients with MR by pulsed Doppler and in an additional 15 of 16 patients whose MR was demonstrated by color flow Doppler. There were no false positives (sensitivity 54%, specificity 100% for both). The ratio of single-plane, maximal jet area to left atrial area was used to grade MR severity with mild defined as 40%. Cine MRI classification was identical to pulsed Doppler echocardiography in 26 of 30 patients and to color flow Doppler in 16 of 20 patients with no differences of >1 grade. Cine MRI consistently depicted smaller flow disturbances than pulsed Doppler (slope = 0.65) or color flow Doppler (slope = 0.60). Nonetheless, the tine MRI area ratio correlated well with pulsed Doppler (r = 0.78) and with color flow Doppler (r = 0.74). Thus, planar analysis of tine MRI in patients with MR of varying severity gave results that were similar to Doppler echocardiography. At present, for routine clinical assessment of MR, the benefits of tine MRI may be limited to patients in whom transthoracic Doppler echocardiography is not adequate. (AmJ Cardiol 1990;66:621-625)
From the Cardiovascular Section, Department of Medicine and the Devon Imaging Center, the Department of Radiology. Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania. This study was supported in part by the Commonwealth of Pennsylvania Health Services Contract and by a Fellowship Grant from the Southeastern Pennsylvania Chapter of the American Heart Association, Philadelphia, Pennsylvania. Dr. Aurigemma is now at the Department of Medicine, Division of Cardiology, University of Massachusetts Medical Center, Worcester, Massachusetts; Dr. Schiebler is now at the Department of Radiology, University of North Carolina Medical Center. Chapel Hill, North Carolina. Manuscript received April 10, 1990; revised manuscript received and accepted April 30, 1990. Address for reprints: Gerard Aurigemma. MD, Division of Cardiology. University of Massachusetts Medical Center, 55 Lake Avenue North, Worcester. Massachusetts 01655.
ine magnetic resonance imaging (MRI) uses gradient refocused echoes to produce dynamic, high-resolution tomographic images of cardiac motion and blood flow without contrast administration. I The heart can be imaged at multiple levels simultaneously and the resulting images displayed in tine loop fashion. Moreover, turbulent flow associatedwith valvular regurgitation appearsas a discrete signal void2,3and is readily distinguishable from laminar flow. Pulsed and color flow Doppler echocardiography are used clinically for the evaluation of mitral regurgitation (MR), but can sample only selectedplanes through the heart4 Because tine MRI can tomographically sample the entire cardiac volume and distinguish turbulent from laminar flow, it may be valuable to apply this technique to the study of MR. Our study assessedthe ability of tine MRI to detect MR and to quantitate the size of the associated flow disturbance as compared to pulsed and color flow Doppler.
C
METHODS Patients:
Forty patients aged 25 to 88 years with a medical history of valvular or coronary artery disease and 10 volunteers aged 25 to 70 years without known heart diseasemade up the study group. Patients with known valvular heart diseasewere approached for participation in the study based on results of a recent Doppler echocardiographic examination or at the suggestion of their physician. In addition, patients with clinically normal valves who had a tine MRI examination for evaluation of coronary artery bypass graft patency5 were also included in the study group. All subjects were clinically stable, in regular sinus rhythm and gave informed consent. The study was approved by the institutional review board of the Hospital of the University of Pennsylvania. Cine magnetic resonance imaging technique: MRI was performed in a 1.5 Tesla Signa (General Electric) MRI system. Each study began with electrocardiographically-gated multislice coronal spin echo images of the thorax for cardiac localization. The image matrix was 128 X 256, interpolated to 256 X 256 for display and the field of view was 32 or 40 cm depending on patient size. The coronal serieswas used to define the extent of the left atrium. At least 2 tine MRI seriesof two to four 5 to 10 mm-thick axial slicesthrough the left atrium and left ventricle with interslice distances of 5 mm to 2 cm were obtained in all subjects.Early in the series,however, we concluded that the optimal imaging approach consisted of interleaving 3 sets of three 10
THE AMERICAN
JOURNAL
OF CARDIOLOGY
SEPTEMBER
1, 1990
621
mm-thick slices with an interslice distance of 20 mm per series; this approach was used routinely for the patients in the remainder of the series. The tine MRI techniques used flip angles of 30’ and gradient refocused echoes with an echo time of 12.5 ms and repetition time of 22 ms. The radio frequency pulse, readout gradient, phase-encoding gradient and slice-selective gradient occurred independently of the cardiac cycle every 22 ms. The R wave of each electrocardiographic cycle controlled advancement of the phase-encoding gradient to the next image line. In this study, frame rate was determined by dividing the RR interval by the repetition time and then by the number of slices desired in the tine loop. Thus, for an RR interval of 800 ms (heart rate 75 beats/mm) 36 frames were available. If 3 slices were desired, each tine loop contained 12 frames per cardiac cycle. The number of frames per cardiac cycle for subjects in this series ranged from 5 to 26. It is also possible to obtain more frames by interpolation in the image reconstruction. Two to 4 signal averages per image series were obtained to optimize image quality. Each acquisition took 128 cardiac cycles, 1 for each Y line in the image. Time per acquisition at the time of this study included 2 to 3 minutes for prescan, 3 to 5 minutes for imaging and 5 to 10 minutes for imaging processing. Total time per study was approximately 50 minutes. Subsequent software improvements permit patient time in the system of roughly 35 minutes with image reconstruction time of 3 minutes per series.
Doppler techniques: Thirty subjects were evaluated using standard 2-dimensionally guided pulsed Doppler mapping of MR using commercially available systems. A 2.5 MHz Doppler flow probe was used to perform examinations from apical and parasternal windows. A point-by-point meticulous search for disturbed flow was conducted from apical 2- and 4-chamber and parasternal long-axis views in all patients. Twenty subjects were evaluated using color flow Doppler mapping with a Hewlett-Packard model 77020 system. A careful search was made in each view to record the maximum systolic left atria1 area of disturbed flow. Evaluatiin of rewlts--eine imaging: Qualitative recognition
magnetic
resonance
of MR by tine MRI was arrived at by consensus at a reading session attended by 2 cardiologists and 2 radiologists. Jets of MR were defined as dynamic zones of signal void in the left atrium in systole extending from the mitral valve plane. For each study, the tine MRI frame showing the maximal signal void associated with MR was selected. Quantitation was performed by digitizing the area of the signal void and the left atria1 area from film hard copy using a microcomputer equipped with a backlit, transparent high-resolution digitizer (Hewlett-Packard 9825A). Because the imaging planes were different by tine MRI and Doppler echocardiography, the ratio of maximal jet area to left atria1 area was used to normalize results. Severity was graded as follows: mild, jet area to left atria1 area ratio lgl-ade.
622
THE AMERICAN
JOURNAL
as an area ratio as HO%. Classifi-
OF CARDIOLOGY
VOLUME
66
isunifOrmhighsignalthfw&fd all 4 cardiac chambers and etructurefi have attemmtin the denceding aorta. Myocardial ed signal compared to bbod pool. Moderator band is visualiz0d in right venbide. B, arrow points to dosed mitral valve. C and D, in later systole, uniform high signal persists in lett atrium and systolic s~pt~l thickening is also shown. A = descendingaorb;LA=leftatriun;LV=leRventride;RA=ri%lt atriun;RV=rightventdcle.
area to left atria1 area greater than 40%, after the work RESULTS of Helmcke et al.” Mitral regurgitation was detected by Doppler echoDoppler echocardiography: Qualitative recognition cardiography in 32 of 50 subjects and by tine MRI in of MR by Doppler echocardiography was determined 30 of 50 subjects. Mitral regurgitation was found in 16 by 2 observers who were unaware of the tine MRI re- of the 20 subjects examined by color flow Doppler and sults. A single observer, blinded to the tine MRI results, 16 of 30 examined by pulsed Doppler (Figure 1). Cine assessed the severity of MR. Mitral regurgitation was MRI detected MR in 15 of 16 subjects with MR by defined as a broad high-velocity spectrum with aliasing pulsed Doppler and 15 of 16 with MR by color flow detected in the left atrium and extending from the valve Doppler (Figure 1). There were no false positives. Overplane in systole for pulsed Doppler. Using color flow all sensitivity was 94% and specificity was 100% with Doppler, MR was defined as a retrograde flow region Doppler flow mapping as a reference standard. with or without variance and aliasing extending from Using tine MRI, intracardiac laminar blood flow the mitral valve into the left atrium in systole and con- generates a higher signal than myocardial structures firmed by pulsed Doppler. The area of the MR color (Figure 2). A regurgitant jet of MR was visualized as a flow disturbance was outlined on clear plastic film and signal void in the left atrium in systole (Figure 3). The digitized. In subjects who had pulsed mapping alone, a extent, shape and direction of the signal void varied perimeter of the zone in the left atrium where turbulent from patient to patient. Also, there often was dynamic flow was detected was made on a clear plastic overlay variation in the signal void size and shape in a single and digitized. The jet area to left atria1 area was ex- plane throughout systole (Figure 4). pressed as a ratio. The severity of MR was graded as Classification: Cine MRI classification was identical mild if the ratio was 40%.” tical to color flow Doppler in 16 of 20 (80%) subjects. There were no differences of >l grade. Statistical analysis: For the purposes of this study, Doppler echocardiography was taken as the reference Jet area/left atrial area comparison: Figure 5 shows standard. Sensitivity and specificity of tine MRI com- % left atrium occupied by the MR jet by color flow pared to color flow Doppler and pulsed Doppler were Doppler and tine MRI. The regression equation was calculated. The classification of MR as absent, mild, tine MRI = 3.8 + 0.60 color Doppler (r = 0.74). For moderate or severe by tine MRI and pulsed and color pulsed Doppler, the relation was tine MRI = 4.6 + flow Doppler on single-plane maximum jet area left 0.65 pulsed Doppler (r = 0.78) and is shown in Figure atria1 area ratio was compared to define semiqualitative 6. Cine MRI, using the technique described, therefore agreement. In addition, absolute jet area/left atria1 cor- consistently depicted a smaller relative area of flow disrelations were compared by linear regression.
FlGURE 3. Axial tine magnetic remname images taken from the study of a patient with moderate mitral regurgitation. Grientation is similar to Figure 2 for these systolic images. A, /afge afrow paints to wedge-shaped signal void in left atrium dire-&d post&&y. Small arrows point to closed mitral valve leaflets. B, C, D, shape of signal void does not appear to change significantly in these systolic images.
FIGURE 4. Axial systolii tine magnetic resonance images from study of a patient with severe mitral regurgitath. A, /afge whife arrow points to signal void in left atrium emanating from region of dosed mitral kafkts (sma// arrows). Et, C, D, area of signal void enlarges and propagates medially and postthroughout systolo. In C and D, signal void has diminished in intensity but appears to occupy a larger perceh age of the left atria1 area, including pulmonary veins.
THE AMERICAN
JOURNAL
OF CARDIOLOGY
SEPTEMBER
1, 1990
623
turbance when compared to pulsed or color flow Doppler mapping. DISCUSSION
True volumetric estimation of the severity of valvular regurgitation can only be accomplished by comparing forward left ventricular stroke volume with total left ventricular stroke volume. Such estimates of regurgitant volume are only meaningful in patients with single-valve regurgitation. While the clinically accepted method for computing regurgitant volume subtracts Fick (or thermcdilution) cardiac output from angiographic cardiac output,6 regurgitant volume has also been estimated noninvasively by radionuclide angiography7 and by tine computed tomography.8 In addition, 2-dimensional9 and Doppler echocardiographytO have been used to compare right and left ventricular stroke volumes and to estimate regurgitant fraction. However, for MR, clinicians commonly rely on semiquantitative assessment of the area of flow disturbance, such as is provided by Doppler echocardiography or by angiographic (1 +, 2+) grading. In general, there is a good correlation between Dopplerderived regurgitant jet area and qualitative angiographic assessment of regurgitation.4 Because tine MRI, like color flow Doppler mapping, displays disturbed cardiac flow, it appears to be a suitable method for the semiquantitative grading of MR. Cine magnetic
resonance
imaging
signal
properties:
With gradient refocused imaging, myocardial structures 50
cMR1= 3.6 t 0.60 CFD r=.74 40
have attenuated signal relative to blood because their less mobile protons remain in the imaging plane and are continuously excited. The signal intensity of flowing blood depends, in part, on its velocity and on whether the flow is laminar or turbulent. In general, laminar flowing blood generates uniform, bright signal; this may be due to the continuous entry into the imaging plane of fresh blood whose spins are “unsaturated” and give maximal signal.tJ’ Factors that may also contribute to the high signal are the high-proton density of blood2 and the lack of phase cancellation induced signal loss and short echo times. The regurgitant jet of MR was associated with loss of signal in a region of the left atrium in systole. Multiple mechanisms of signal reduction may exist. Dephasing of spins caused by shear between flow layers moving at different velocities adjacent to heart or blood vessel wall can produce a signal void. Time of flight effects are thought to be minimal for tine MRI. Fram et all2 have shown the velocity dependence of signal amplitude at short repetition times in an in vitro model with evidence of a decrease in signal at higher velocities. For regurgitant jets, signal voids are presumably due to turbulence13; phase cancellation of signal may occur among protons moving at different velocities.2 Cine magnetic resonance imaging/Doppler comparisons: Overall qualitative correlation between tine MRI
and Doppler echocardiography was good with no differences of > 1 grade. The quantitative correlation was also good, though tine MRI appeared to show smaller flow disturbances that pulsed Doppler (slope = 0.65) or color flow mapping (slope = 0.60). This discrepancy may be due to multiple factors. First, temporal resolution of
.
. .
60 .
cMR1= 4.6 t 0.65 PO i-s.78
50
10
0
0
I
10
.
I . I 20 30
I ’ 40 50
0 3
PULSED DOPPLER
COLORFLOWDOPPLER FIGURE 5. Comparison between jet areahefi atrlal area ratio (%LA) derived from cdor Raw Doppler (CFD, x-axis) ad from mmaance Imaging (cMRI, y-axis). cMRl = tine einemagnetk-luaimaging.
624
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
66
FIGURE
6. Comparison
between
jel area/left
atrial
area ratio
derived lmm pulsed Doppler (x-axis) and from tine magnetk resonance imaging nance imaging.
(cMRI,
y-axis).cMRI
= tine magnetic
reso-
mapping was different for tine MRI than for pulsed Doppler. Thus, the maximal jet area defined for pulsed Doppler was mapped throughout systole and the composite of all points positive for MR were taken. Similarly, for color flow Doppler, the frame rates ranged from 15 to 30/second. In contrast, as noted before, tine MRI has an effective frame rate of only 12/cardiac cycle/ slice in a 3-slice acquisition when the heart rate is 75 beats/min. Because MR flow varies over time, the differing temporal resolution of sampling by Doppler echocardiography could result in larger apparent jet area even if both Doppler echocardiography and tine MRI jets were actually identical. Secondly, differences in imaging planes used could affect the results, because axial tine MRI images lie closer to the short-axis plane of the MR jet, while the apical Doppler window lies closer to the jet’s long-axis plane. Recent software enhancements permit oblique imaging with tine MRI and further studies should permit comparison of comparable image planes. Additionally, while Doppler and tine MRI both detect turbulent flow, they may in fact be measuring different physical phenomena. Finally, discrepancies between tine MRI and Doppler grading of MR may have been due to differences in loading conditions between the time of the 2 studies. Patients had their tine MRI and Doppler studies on the same day whenever possible. Nonetheless, because hemodynamic monitoring was not performed during MRI, hemodynamic changes, with attendant effects on regurgitant jet area, may have taken place. Study limitations: There are a number of important limitations to this study. The size of the color flow Doppler jet of MR has been shown to vary depending on technical factors, including gain adjustment and filter settings.4,14Technical factors also influence the size of the flow disturbance as depicted by tine MRI; the size of the signal void associatedwith valvular regurgitation appears to be dependent on echo time. In addition, as noted before, tine MRI frame rate is dependent on the patient’s heart rate and number of slices per acquisition. Finally, window and threshold settings for the tine MRI display were set to provide maximum contrast for each image; no attempt was made to standardize these settings. This study may also be criticized for the use of Doppler echocardiography as a reference standard. The limitations of Doppler semiquantitative grading have been well described.15However, we chose to compare the results of tine MRI grading to Doppler echocardi-
ography becausethe latter remains the standard noninvasive method used to evaluate valvular regurgitation. Acknowledgment: We wish to thank Kathleen Bogin, RN, and Margaret Dunlop, RN, for their patient recruitment and data management, Robert Morrow for his assistancein processingthe Doppler data, the technical staff of the Devon Center for their assistance,and Janine Carter and Beverly Butler for their help in the preparation of the manuscript.
REFERENCES 1. Sechtem U, Pllugfelder PW. White RD, Gould R, Halt W, Lipton M, Higgins C. Cine MR Imaging: potential for the evaluation of cardiovascular function. Am J Kudiology 1987;I 48:239-246. 2. Schiebler M, Axcl L, Reichek N, Aurigemma G, Yeager B, Douglas P. Bogin K. Kreaael H Correlation of&x MR imaging with 2-dimensional pulsed Doppler echocardiography in valvular inaufliciency. J Compur .4ssisr Tomogr /987;/ I627-632. 3. Sechtem U, Pllugfelder PW, Cassidy MM. White R, Cheitlin M, Schiller N, Higgins C. Mitral or nortic regurgitatmn: quantification of regurgiwnt volumes with tine MR imaging. Radrology /988:/67.425-430. 4. Helmcke F. Nanda NC, Hsiung MC, Soto B, Ade) C. Goyal R, Gatewood R. Color Doppler aaaessment of mitral regurgitation with orthogonal planes. Circulalion l987;75.175~lU3. 5. Aurigemma G, Relchek N, Axcl L, Schiebler M, Ilarris C, Kressel H. Nomn\aswedeterminatlon ofcoronary artery bypass graft patency bq tine MRI. Circulution 1989,80:1595-1602. 6. Croft CH, Lipscomb K, Mathia K. Firth B, Nlcod P. Tilton G, Winniford M, Hillis L. Lmitations of qualitative angiographx grading in aortic or rmtral regurgitation Am J Cardiol 1984;S3:1593-1598. 7. Kelback H. Alderahvile J, Svcndscn JH, Folke K, Nielsen SL, Munck 0. Combined first pars and equilibrium radionuchde cardiographic determination of stroke volume for quantitation of valvular regurgitation J Am Co// Curd;01 I9XX:I I ,769 773 8. Rater S, Rumbcrgcr .I. Feiring A. Stanford W, Marcus M Prccislon measuremen& or right and left ventricular volume by tine CT. Circulution /986:74:890900. 9. Biumlein S, Bouchxd A, Schdler NB, Dac %I, Bird B. Ports T , Botvimk E. Quantltation of mitral regurgitation by Doppler cchocardlography. Circu/a~io,r 1986:74.306 314. 10. Rokcy R, Sterling K. Logbhi W. Sartorl M. L.lmacher M. Kuo L. Quinones M. Determination of regurgitant fraction m wlated mitral oraortic regurgitation by pulsed Doppler two-dm~cn&~nal cchocardlography. J .4n1 (bll C’ardrol /9X6,7.1?73 127x. 11. Aurlgemma GP, Reichek N, AxeI I., Kreascl H. Cudiac tine MRI: normal Intracardiac and great VCS~CI Ilox. C/in KG IY87.35.25YA 12. Fran E. Hedlund I.. Dlrnlck R, Glover G, Herlkens R. Parameters dctermining the signal of flowing fluid in gradient refocused imagmg: flow velocity, TR. and nip angle. Prow~d~n~s of thr Soort~~fur Ma~nrri~~ Krsotronce vz Mrdrcinp Vth Anma/ Mrctin~, /Y&5. 1.84. 13. Evans A. Blmder R, Her&ens R. Sprit/cr C, Kuethe D, Fran1 E, Hedlund L. F.ffccta of turbulence on signal intenslt) in gradxnt echo mugcs. Iwe~t Radio/ lYXX.23.517 51x. 14. Wang M, Matsut~~ura M. Surukl h, Omoto R. Tcchnicul and biologic wurcc~ ofvariability III the mapping of axtIc. mitral and tricuspid color now jets. .4nt J Cbrdrol 1987;60.-847 851 15. Bolgcr AF, Eiglcr %L. hlaurer G Qudllt~f)~llg\;il\ulllr regurgltatmn: limltattons and inherent ;wumptwnruf Doppler tcchniqucr. (;r
