MESA Study: MRI Evaluation

MESA Study: MRI Evaluation

1. Summary

2. Cardiac Evaluation
    Left Ventricle
    Right Ventricle
    Atrial Assessment
    Left Atrium
    Right Atrium
    Myocardial function: MR Tagging
    Cardiac Perfusion

3. Aortic Evaluation
    Stroke Volume
    Wall Shear stress rate
    Aortic Compliance
    Vascular Impedance
    Plaque Characterization

4. MR Imaging Protocol

5. References

1. Summary

Objectives for the MRI examination include 1) obtaining reproducible, high quality data suitable for the MESA epidemiological study, 2) using state of the art methods for analysis and data collection, but requiring methods that have been validated and that are not proprietary, and 3) specifying protocols that may be performed at all Field Centers. In addition, as originally specified, the MR evaluated is limited to a total of 30 minutes of participant time. Finally, because of the high volume of examinations and funding restrictions, complete MR quantitative analysis must be completed in approximately 30 minutes of time.

2. Cardiac Evaluation

Left ventricle.

Left ventricular volume is an important diagnostic and prognostic factor after myocardial infarction and in patients with valvular abnormalities (1), and essentially all indices of left ventricular function derive from measurements of volume and pressure (2). Current approaches to patient therapy depend on knowledge of systolic performance of the left ventricle (3). Ejection fraction and cardiac output measures, although dependent on afterload, give an overall index of global ventricular performance (2). Left ventricular hypertrophy and mass are strong predictors for cardiovascular disease morbidity or mortality (4-7). However, measures of global left ventricular function using echocardiography or ventriculography depend on geometric assumptions that are inaccurate in patients whose ventricles do not conform to ideal geometric models. MRI methods use Simpson's rule and therefore avoid introduction of geometric assumptions that lead to inaccuracy. MR evaluation of cardiac chamber dimensions is facilitated by the ability to obtain images with good spatial resolution in any tomographic plane. Thus, errors associated with geometric assumptions, as is the case with echocardiography, are avoided (8).

Accurate left ventricular volume measurements may be obtained with spin-echo and conventional cine MRI techniques (9), as well as with the proposed breath-hold segmented k-space cine acquisitions (10). Compared to cast volume displacement of the left ventricle, a correlation of 0.99 was found, with standard error of the estimate (SEE) of 4.9 cm3(11). Interstudy variability has been determined to be less than 4-6% for cine MRI assessment of LV mass (9, 12, 13). Standard deviation of MRI determined left ventricle ejection fraction is 4% (9). Cardiac index and end systolic wall stress interstudy variability is slightly greater at 8-10%, likely due to physiologic variation in afterload (9). The above referenced studies were performed by a single site. The multi-site MESA trial should expect variabilites to increase by 50% to 100%, ie. LV mass variability in a multi-study trial is likely 7-10%/

MR can measure parameters of cardiac function more accurately and reproducibly than echocardiography and these global variables have been explored previously using echocardiography in large epidemiological studies. Therefore, we specifically recommend that dynamic measures of cardiac contraction and of end diastole be measured, thus taking full advantage of MRI's unique strengths.

Diastolic function has been recognized recently as becoming abnormal before systolic function and may therefore be useful in the assessment of subclinical disease. Diastolic dysfunction is felt to be a common problem in older people, ultimately leading to heart failure in an estimated 40% of older persons. Isolated diastolic dysfunction causes left atrial hypertension, pulmonary venous hypertension, and congestion. Key parameters that can be readily analyzed using conventional (nontagged) MR images include peak filling rate, and time to peak filling rate of the left ventricle. In addition, the peak contraction rate and time to peak contraction is also readily estimated using the MASS software program. This method calculate parameters of cardiac function by analysis of all cardiac phase information (~10 slice locations x approximately 20 cardiac phases=200 images.

(return to top)

Right Ventricle.

Right ventricular functional assessment is complex due to the unusual shape of that cavity (14). Echocardiography is of limited value in assessing the right ventricle due to difficulties with visualizing the endocardial borders and the complex geometry (15). MRI assessment of the right ventricle is based on Simpson's rule, and, like the left ventricle assessment, does not require geometric assumptions.

MR validation data for the right ventricle are less extensive than for the left ventricle (16). Excellent correlation, however, has been found between MRI and indicator dilution methods (r = 0.93) and ventriculography (r = 0.96) (15, 17). In normal individuals, right ventricular stroke volume has been shown to be equal to left ventricular stroke volume (15, 18). Impairment of right ventricular diastolic function has been suggested to be a more sensitive marker of myocardial disease than right ventricular ejection fraction (16). In patients with dilated cardiomyopathy, although ejection fraction was normal, time to peak filling rate was prolonged and filling fraction of the right ventricle was reduced (19). Thus, right ventricular function should be explored as a marker of subclinical disease. Right ventricular mass allows detection of early right ventricular hypertrophy (20). Interobserver and intraobserver variation of right ventricle ejection fraction by MRI in one study was approximately 6% using Simpson's rule for cavity size measurement (21). Multi-institutional studies are likely to be twice this value.
(return to top)

Atrial assessment.

Atrial assessment by MRI showed 7% underestimation of cavity dimension compared to atrial casts. Information regarding reproducibility of measurements and intra- and inter-observer variability has not been specifically studied or described in the literature (22).

Left atrium.

Left atrial function has largely been ignored in population-based studies due to difficulty in measurement using echocardiography or other noninvasive methods. As a compliant reservoir for delivery of blood to the ventricle, atrial function augments ventricular function (23, 24). Due to the marked volume response of the atria to pressure changes, atrial volume may potentially be an early indicator of subclinical disease. Left atrial analysis from MRI data is based on end systolic and end diastolic images, similar to the right ventricle.
Right atrium.
Right atrial assessment was requested in RFP HC-98-11. However, we believe that the additional time required to perform these measurements is unwarranted due to lack of prior studies indicating its relevance to subclinical cardiovascular disease.
(return to top)

Myocardial Function: Cardiac Tagging Methods.

Myocardial tagging using MRI was developed by Zerhouni (25) and extended by Axel and Dougherty (26). These methods have added important insights into regional myocardial mechanics. In contrast to global measures of systolic function such as ejection fraction, MRI myocardial tagging makes it possible to quantify the severity and extent of subtle regional heart wall motion abnormalities, and is therefore relevant to subclinical detection of cardiovascular disease.

The fundamental improvements in regional function assessment using MR tagging techniques rather than echocardiography and non-tagged MRI are two-fold: 1) the same volume of myocardium can be tracked over the heart cycle in order to map the function in a specific region, and 2) very precise quantitative estimates of myocardial shortening and wall thickening can be computed from the images. With myocardial tagging techniques, the normal contraction patterns of the human LV are very consistent from heart to heart (27) and regional mechanical behavior after infarction (28) and during ischemia (29) have been characterized. Patients on day 5 after a first anterior MI with single-vessel disease demonstrate reduced intramyocardial circumferential shortening throughout the LV, including remote noninfarcted regions (30). These observations were relatively subtle and required quantitative techniques to reach a significant conclusion. Quantitative functional analysis by Kramer et. al. (31) showed that angiotensin-converting enzyme inhibition leads both to a reduction in left ventricular remodeling and limits the decline in the function of the adjacent noninfarcted region.

Tag detection is extremely precise and accurate (32-35). Reproducibility is maintained by consistent adherence to acquisition methods and analysis techniques that utilize intrinsic cardiac landmarks. The position of a myocardial tag can be estimated to within 0.1 pixels, corresponding to tag localization to within approximately 150 µm. The reproducibility of tagging was shown by Sayad et al.: normal volunteers were imaged once, then had a repeat scan the same day, and again 4 weeks later. The mean differences in end-diastolic wall thickness, end-systolic wall thickness, and percent wall thickening mean between the baseline and 4-week studies were -0.05±0.7 mm, 0.0±0.9 mm, 0.98%±17%, respectively (36). Correlation coefficients for wall thickness measures performed on the same day and repeated 4 weeks later were 0.94 and 0.95. Although tagging allows the full three dimensional displacement field over the entire left ventricle (27, 28, 37, 38) to be calculated, currently 2D analysis in the circumferential and radial directions is more practical since the analysis is much more rapid. Inclusion of MR tagging images may add substantial power for cardiac analysis relative to conventional echocardiography or MR cine imaging. This is recommended as a substudy due to time constraints.
(return to top)

Cardiac perfusion.

MRI assessment of cardiac perfusion will substantially evolve during the course of the study and become a standard component of MR cardiac analysis. However, in the presence of subclinical cardiovascular disease, it is unlikely that resting, or nonpharmacologic perfusion patterns would be abnormal. Therefore, any such assessment of cardiac perfusion would be more suited to individual site substudy.
(return to top)

3. Aortic Evaluation

Stroke Volume.

Stroke volume and cardiac output may be assessed by cine phase contrast imaging (39, 40) in the aorta, at the level of the pulmonary artery bifurcation using breath-hold techniques (41). Signal intensity is mapped to velocity-induced phase shifts of moving spins in the presence of a magnetic field gradient. Thus, phase shift is proportional to velocity. Stroke volume is derived from velocity and cross-sectional area of the aorta, and cardiac output is obtained from stroke volume multiplied by the average heart rate. These aortic measurements improve accuracy compared to stroke volume/cardiac output by LV volumetric methods if mitral regurgitation or intracardiac shunt is present. Aortic measurements, however, do not account for the fraction of cardiac output that goes to the coronary arteries.
(return to top)

Wall shear stress rate.

The formation of atherosclerotic plaque on the aortic wall is spatially correlated with regions of low and oscillatory wall shear stress. The morphology of the vascular wall also changes in response to hemodynamic factors. In high shear regions, vascular epithelial cells tend to elongate, aligning themselves in the direction of flow while in low shear or oscillatory regions, epithelial cells lack alignment and appear more rounded. This rounded shape pattern seems to contribute to a higher cell turnover rate as well as to an increased permeability to macromolecules. Since geometry and flow patterns are major factors in the hemodynamic conditions present in the artery, and these factors vary significantly among individuals, a noninvasive technique for spatially localized measurements of shear stress or shear rate of the aortic wall could provide a means of designating regions of the aorta that have a propensity for atherosclerotic change.

Most experience with MRI evaluation of aortic wall shear stress rate has focused on the abdominal aorta; atherosclerotic changes are unusual in the ascending aorta (42). Mostbeck et. al. used MR cine phase contrast imaging to look at retrograde flow in the abdominal aorta and found it to be maximal just distal to the renal arteries, and that in this infrarenal region, the posterior wall experiences the longest period of retrograde flow (43). There was a great deal of inter- and intra-individual variability in these studies, however. Oshinski et al. investigated the use of this type of MR data to measure wall shear stress in vivo (44). In eight patients, noninvasive MR-based estimates of shear stress in the infrarenal aorta were significantly lower than those in the suprarenal aorta, and values on the posterior wall of both sections were lower than the inferior wall at the respective locations. Shear stress was calculated as the product of blood viscosity and the velocity gradient at the wall. The velocity gradient was estimated from the velocity images using the edge pixels and the two or three pixels radially inward from the wall. The accuracy of these measurements is highly dependent on the estimated intrapixel location of the arterial wall. The error ranged from -45% to 32% in these experiments based on validation with an analytical model of pulsatile flow. Due to these considerations, wall shear stress calculations are not recommended. It is noted, however, that as analysis routines become less operator-dependent, the same data acquired for the aorta (discussed above) may be useful for wall shear stress analysis, e.g., in a substudy.
(return to top)

Aortic Compliance.

Aortic compliance, or distensibility, can be derived from the pulse wave velocity in the aorta determined by MRI. Distensibility is defined as the ratio of fractional change in cross sectional area to the corresponding change in pressure and is an important component of left ventricular afterload. Low aortic distensibility has been reported in patients with Marfan syndrome, coronary artery disease, and hypertension (45, 46). Additionally, reduced distensibility may be an early indicator of early atherosclerosis (47, 48). Chronic increased vascular loading may lead to left ventricular hypertrophy. Additionally, it may cause or contribute to the development of heart failure, while changes in the material properties of the aortic wall and accompanying aneurysmal dilation may lead to sudden aortic dissection or rupture.

Aortic distensibility may be calculated as (maximum cross-sectional area of the aorta - minimum area)/minimum area x pulse pressure, based on a single level cine MRI acquisition and measurement of brachial artery pressure (49, 50). Aside from the problems of using externally derived cuff pressures, the accuracy of these measurements is extremely sensitive to accurate detection of vessel boundaries, and requires good quality images with high spatial resolution (51, 52). Forbat et al. found that short-term reproducibility studies had a measurement variance of 5% when a 30cm FOV and a 256x192 matrix was used with a spin-echo sequence (51). Gradient echo sequences produced higher measurement variance due to lower vessel wall contrast. Estimates for the MESA trial are approximately 10%.
(return to top)

Vascular impedance.

There are several types of impedance values that have been applied to the vasculature, such as input impedance and characteristic impedance. These values describe the relationships, typically complex, between pressure and flow in a vessel. Although MR has been used to measure distensibility, a component of vascular impedance, to our knowledge it has not been used either to directly measure or estimate vascular impedance per se.

RFP HC-98-11 has related vascular impedance to patterns of flow: abnormal flow patterns can indicate locations in a vessel where vascular impedance changes. An example of this would be the turbulence and wave reflections associated with a stenosis or aortic coarctation. MR can map flow patterns in the aorta using phase contrast techniques to assess flow velocity. In the ascending aorta, MR has been used to study retrograde flow, which has been deemed important for coronary perfusion (53), and to evaluate the complex patterns of flow that occur with aortic stenosis and regurgitation (54). Mostbeck et. al. used a cine MR phase contrast sequence to acquire velocity encoded cross sectional images of the descending aorta (43). Indices of the ratio of retrograde to anterograde flow were formulated, but this work is very preliminary and not currently recommended for the MESA study.
(return to top)

Plaque characterization.

The simplest goal of atherosclerotic plaque characterization would be to differentiate "complicated" plaque with ruptured fibrous caps, intramural hemorrhage and atheromatous debris from uniform, or stable, plaque. Complicated plaques are found 2-6 times more frequently in symptomatic patients, compared to asymptomatic patients, with carotid artery disease. To date, human in vivo plaque characterization has been assessed for the carotid artery (for example, see (55)), optimally using small surface coils applied to the skin surface. Human atheroma has also been evaluated using MRI with in vitro specimens, using T1, T2 and gradient echo images. Long imaging times using spin echo techniques (6-7 minutes) limit applicability in the aorta due to patient motion and aortic pulsation during image acquisition. Specialized pulse sequences for lipid detection (chemical shift imaging, stimulated-echo diffusion-weighted images) may potentially be applied to characterize plaque (56, 57).

In vivo plaque characterization of the human thoracic aorta has not been studied with pathologic confirmation using MRI with surface coils. Thoracic aorta plaque characterization requires further study before implementation in the MESA investigation. Important measures, such as the extent of lipid accumulation, could be rapidly assessed by chemical shift breath-hold imaging. However, validation studies of the technique are preliminary and accuracy is unknown. Potentially, the MESA trial could evaluate carotid plaque composition using dedicated high resolution surface coils, subject to consideration of participant's time constraints and funding.

This technique of carotid imaging is completely different than the MR methods used for heart image acquisition. Carotid plaque imaging cannot simply be "added-on" while the MESA study participant is in the MR scan area. It involves repositioning of the patient and coils, and would approximately double the participant's time and the MR study cost. Techniques for analysis have not been widely performed, nor reproducibility established. The optimum data for SCD coding has not been determined. Carotid plaque MR imaging appears to be an exciting new area of investigation that we will continue to explore. For the MESA study, the Steering Committee should recognize that plaque characterization represents a new, and additional test, much like electron beam CT is a different test than cardiac MRI. The technical aspects of carotid image evaluation, likewise, is almost completely unrelated to the analysis of cine cardiac images. The CHS study is currently performing a pilot study of 32 patients to analyze carotid MRI. The MESA MR Committee will monitor the results and success of this pilot study.
(return to top)

4. MR Imaging Protocol

Series	Description	MR sequence parameters	Measured Parameter	Software Tool
1	Localizer, coronal (1 breath-hold)	TR/TE/FA 6/min/20° 1 NEX, gated, NVS=32, 256x160, 3/4 FOV, 40 cm- adjust to body size, 8x2mm, 4 slices flow comp, 32 kHz
2	Pseudo long axis, cine (1 breath-hold)	TR/TE/FA 6/min/20° 1 NEX, gated, NVS=32, 256x160, 3/4 FOV, 34 cm- adjust to body size, 8x2mm, 4 slices flow comp, 32 kHz	LA ED/ES volume	MASS
3	Short axis, cine (8 breath-hold)	TR/TE/FA 6/min/20°, 1 NEX, gated, NVS=8, 256x160, 3/4 FOV, 34 cm - adjust to body size 8x2mm, flow comp, 32 kHz	RV ED/ES volume LV ED/ES volume Ejection fraction Cardiac output LV ED thickness Time to max sys. shortening Rate of max sys. shortening Time to max dias. relax. Rate of max dias relax. LV mass	MASS MASS MASS MASS MASS MASS MASS MASS MASS MASS
4	Horizontal long axis, cine (1 breath-hold)	TR/TE/FA 6/min/20°, 1 NEX, gated, black blood NVS=8, 256x160, 3/4 FOV, 34 cm- adjust to body size, 8x4mm, 8 x 2 mm, 32kHz	LA ED/ES volume	MASS
5	Axial cine phase contrast (2 breath-holds)	2D cine phase contrast, TR/TE/FA 10/5/20°, gated, 256x128, 1/2-3/4 FOV, 32 kHz, VENC 180 cm/sec 4 views per segment, 8 mm thickness	Stroke volume Cardiac output Distensibility	FLOW FLOW FLOW

(return to top)

5. References

1. White HD, Norris RM, Brown MA, al. e. Left ventricular end-systolic volume as a major determinate of survival after recovery from myocardial infarction. Circulation 1987; 76:44.

2. Braunwald E. Assessment of cardiac function. In: E. Braunwald, ed. Heart Disease: A Textbook of Cardiovascular Medicine. 4th ed. ed. Philadelphia: W.B. Saunders, 1992: 419.

3. Conway J. Clinical assessment of cardiac output. Eur Heart J 1990; 11 (suppl I):148-152.

4. Koren MJ, Devereux RB, Casale PN, Savage DD, Laragh JH. Relationship of left ventricular mass and geometry to morbidty and mortality in uncomplicated essential hypertension. Ann Intern Med 1991; 114:345-352.

5. Casale PN, Devereux RB, Milner M, al. e. Value of echocardiographic left ventricular mass in predicting cardiovascular morbid events in hypertensive men. Ann Intern Med 1986; 105:173-178.

6. Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham heart study. N Engl J Med 1990; 322:1561-1566.

7. Bikkina M, Levy D, Evans JC, et al. Left ventricular mass and risk of stroke in an elderly cohort. JAMA 1994; 272:33-36.

8. Shapiro EP, Rogers WJ, Beyar R, et al. Determination of left ventricular mass by magnetic resonance imaging in hearts deformed by acute infarction. Circulation 1989; 79:706-711.

9. Pattynama PMT, Lamb HJ, van der Velde EA, van der Wall EE, de Roos A. Left ventricular measurements with cine and spin-echo MR imaging: a study of reproducibility with variance component analysis. Radiology 1993; 187:261-268.

10. Sakuma H, fujita N, Foo TKF, et al. Evaluation of left ventricular volume and mass with breath-hold cine MR imaging. Radiology 1993; 188:377-380.

11. Rehr RB, Malloy CR, Filipchuk NG, Peshock RM. Left ventricular volumes measured by MR imaging. Radiology 1985; 156:717-719.

12. Semelka RC, Tomei E, Wanger S, et al. Interstudy reproducibility of dimensional and functional measurements between cine magnetic resonance studies in the morpholgically abnormal left ventricle. Am Heart J 1990; 119:1367-1373.

13. Semelka RC, Tomei E, Wagner S, et al. Normal left ventricular dimensions and function: interstudy reproducibility of measurements with cine MR imaging. Radiology 1990; 174:763-768.

14. Sacks MS, Chuong CJ, Templeton GH, Peshock R. In Vivo 3-D reconstruction and geometric characterization of the right ventricular free wall. Ann Biomed Eng 1993; 21:263-275.

15. Møgelvang J, Stubgaard M, Thomsen C, Henriksen O. Evaluation of right ventricular volumes measured by magnetic resonance imaging. Eur Heart J 1988; 9:529-533.

16. Pattynama PMT, De Roos A, Van der Wall EE, Van Voorthuisen AE. Evaluation of cardiac function with magnetic resonance imaging. Am Heart J 1994; 128:595-607.

17. Kondo C, Caputo GR, Semelka R, Foster E, Shimakawa A, Higgins CB. Right and left ventricular stoke volume measurements with velocity -encoded cine MR imaging: in vitro and in vivo validation. AJR 1991; 157:9-16.

18. Sechtem U, Pfulgfelder PW, Gould RG, Cassidy MM, Higgins CB. Measurement of right and left ventricular volumes in healthy individuals with cine MR imaging. Radiology 1987; 163:697-702.

19. Suzuki J, Caputo GR, Masui T, Chang JM, O'Sullivan M, Higgins CB. Assessment of right ventricular diastolic and systolic function in patients with dilated cardiomyopathy using cine magnetic resonance imaging. Am Heart J 1991; 122:1035-1040.

20. Pattynama PMT, Willems LNA, Smit AH, E.E. VdW, De Roos A. Early detection of cor pulmonale with MR imaging of the right ventricle. Radiology 1992; 1992:375-379.

21. Niwa K, Uchishiba M, Aotsuka H, et al. Measurement of ventricular volumes by cine magnetic resonance imaging in complex congenital heart disease with morphologically abnormal ventricles. Am Heart J 1996; 131:567-575.

22. Järvinen VM, Kupari MM, Hekali PE, Poutanen V-P. Right atrial MR imaging studies of cadaveric atrial casts and comparison with right and left atrial volumes and function in healthy subjects. Radiology 1994; 191:137-142.

23. Lau V-K, Sagawa K. Model analysis of the contribution of atrial contraction to ventricular filling. Ann Biomed Eng 1979; 7:167-201.

24. Suga H. Importance of atrial compliance in cardiac performance. Circ Res 1974; 35:39-43.

25. Zerhouni EA, Parish DM, Rogers WJ, Yang A, Shapiro EP. Human heart: tagging with MR imaging--a method for noninvasive assessment of myocardial motion. Radiology 1988; 169:59-63.

26. Axel L, Dougherty L. Heart wall motion: improved method of spatial modulation of magnetization for MR imaging. Radiology 1989; 172:349-50.

27. Moore CC, Lugo-Olivieri CH, Lee L, LIma JAC, McVeigh ER, Zerhouni EA. 3D deformation of the normal human left ventricle: assessment by MR tissue tagging. Society of Magnetic Resonance, Book of Abstracts 1995;

28. Lugo-Olivieri CH, Moore CC, Poon EG-C, Lima JAC, McVeigh ER, Zerhouni EA. Temporal evolution of three dimensional deformation in the ischemic human left ventricle: assessment by MR tagging. Society of Magnetic Resonance: Book of Abstracts 1994; 3:1482.

29. Moore CC, McVeigh ER, Mebazaa A, Zerhouni EA. Use of striped tagging to measure three-dimensional endocardial and epicardial deformation throughout the canine left ventricle during ischemia. J Magn Reson Imag 1993; 3P:124.

30. Kramer CM, Lima JA, Reichek N, et al. Regional differences in function within noninfarcted myocardium during left ventricular remodeling. Circulation 1993; 88:1279-88.

31. Kramer CM, Ferrari VA, Rogers WJ, et al. Angiotensin-converting enzyme inhibition limits dysfunction in adjacent noninfarcted regions during left ventricular remodeling. J Am Coll Cardiol 1996; 27:211-217.

32. McVeigh ER, Zerhouni EA. Noninvasive measurement of transmural gradients in myocardial strain with magnetic resonance imaging. Radiology 1991; 180:677-683.

33. Young AA, Axel L, Dougherty L, Bogen DK, Parenteau CS. Validation of tagging with MR imaging to estimate material deformation. Radiology 1993; 188:101-108.

34. Moore CC, Reeder SB, McVeigh ER. Tagged MR imaging in a deforming phantom: photographic validation. Radiology 1994; 190:765-9.

35. Atalar E, McVeigh ER. Optimum tag thickness for the measurement of position with MRI. IEEE Trans Med Imaging 1994; 13:152-160.

36. Sayad D, Willet DW, Chwialkowski M, al. e. Rapid, noninvasive quantitation of regional left ventricular function using gradient-echo MRI with myocardial tagging. Interstudy reproducibility. Am J Cardiol 1995; 76:985.

37. Moore CC, O'Dell WG, McVeigh ER, Zerhouni EA. Calculation of three-dimensional left ventricular strains from biplanar tagged MR images. J Magn Reson Imaging 1992; 2:165-75.

38. O'Dell WG, Moore CC, Hunter WC, Zerhouni EA, McVeigh ER. Three-dimensional myocardial deformations: calculation with displacement field fitting to tagged MR images. Radiology 1995; 195:829-35.

39. Nayler GL, Firmin DN, Longmore DB. Blood flow imaging by c