Measurements of changes in left ventricular volume, strain, and twist during isovolumic relaxation using MRI

June Cheng-Baron, Kelvin Chow, Nee Scze Khoo, Ben T. Esch, Jessica M. Scott, Mark J. Haykowsky, John V. Tyberg, Richard B. Thompson


Left ventricular (LV) active relaxation begins before aortic valve closure and is largely completed during isovolumic relaxation (IVR), before mitral valve opening. During IVR, despite closed mitral and aortic valves, indirect assessments of LV volume have suggested volume increases during this period. The aim of this study is to measure LV volume throughout IVR and to determine the sources of any volume changes. For 10 healthy individuals (26.0 ± 3.8 yr), magnetic resonance imaging was used to measure time courses of LV volume, principal myocardial strains (circumferential, longitudinal, radial), and LV twist. Mitral leaflet motion was observed using echocardiography. During IVR, LV volume measurements showed an apparent increase of 4.6 ± 1.5 ml (5.0 ± 2.0% of the early filling volume change), the LV untwisted by 4.5 ± 1.9° (36.6 ± 18.0% of peak systolic twist), and changes in circumferential, longitudinal, and radial strains were +0.87 ± 0.64%, +0.93 ± 0.57%, and −1.46 ± 1.66% (4.2 ± 3.3%, 5.9 ± 3.3%, and 5.3 ± 7.5% of peak systolic strains), respectively. The apparent changes in volume correlated (P < 0.01) with changes in circumferential, longitudinal, and radial strains (r = 0.86, 0.69, and −0.37, respectively) and untwisting (r = 0.83). The closed mitral valve leaflets were observed to descend into the LV throughout IVR in all subjects in apical four- and three-chamber and parasternal long-axis views by 6.0 ± 3.3, 5.1 ± 2.4, and 2.1 ± 5.0 mm, respectively. In conclusion, LV relaxation during IVR is associated with changes in principal strains and untwisting, which are all correlated with an apparent increase in LV volume. Since closed mitral and aortic valves ensure true isovolumic conditions, the apparent volume change likely reflects expansion of the LV myocardium and the inward bowing of the closed mitral leaflets toward the LV interior.

  • diastolic function
  • active relaxation
  • mitral valve leaflets

diastole is divided into isovolumic relaxation (IVR) and ventricular filling. During IVR, which is defined by mitral and aortic valves being closed, active relaxation of the left ventricle (LV) results in the rapid decline of LV pressure. Although LV volume is commonly assumed to be unchanged during IVR, previous studies have measured changes in LV dimensions during this period. Specifically, increases in LV base-to-apex length have been observed to occur during IVR in dogs (6, 15) and increases in LV circumference or transverse diameter have also been observed during IVR in dogs (29) and sheep (13). In humans, the outward motion of the LV wall has been reported, sometimes with an accompanying, but smaller, inward motion (1, 30). Although these observations were suggestive of volume changes, it is possible that these observations reflected only regional tissue deformation and were not representative of an actual global change in LV volume. Additionally, close examination of published volume-time curves shows apparent volume increases during IVR (5, 7, 8, 12, 32, 37). These volume changes have been proposed to be due to errors in measurement techniques, possibly reflecting regional geometric changes of the LV chamber (7). LV myocardial tissue deformation during IVR has also been measured in terms of circumferential, longitudinal, and radial tissue strains (25, 36) and LV twist (9, 11, 23, 24, 28, 33). The aim of this study is to use magnetic resonance imaging (MRI) to measure LV volume during IVR and to determine its relationship to myocardial tissue deformation measured in terms of principal strains and twist. Since true chamber volume is assumed to be constant during IVR, we hypothesize that imaging-derived volumes increase during IVR and represent myocardial expansion resulting from changes in myocardial tissue strain as the LV relaxes.


Ten healthy subjects (6 males) aged 26.9 ± 3.8 yr were recruited to participate in this study. The protocol was approved by the University of Alberta Health Research Ethics Board, and all participants provided informed consent before participation. The subjects underwent comprehensive cardiac MRI and echocardiography exams. MRI was used to measure time courses of LV volume, circumferential, longitudinal, and radial principal myocardial strains, and twist, whereas echocardiography was used to observe mitral leaflet motion. Study population characteristics are as follows: 55 ± 6 beats/min heart rate, 117 ± 9 mmHg systolic blood pressure, 70 ± 6 mmHg diastolic blood pressure, 196 ± 32 ml end-diastolic LV volume, 77 ± 14 ml end-systolic LV volume, 119 ± 20 ml stroke volume, 61 ± 3% ejection fraction, and 114 ± 21 g LV mass.

Defining LV volume.

Although LV volume must be constant during IVR when it is defined by closed mitral and aortic valves, commonly used methods for volume determination use only endocardial borders, as shown in Fig. 1, and are thus not constrained by the closed heart valves. A plane is used to define the extent of the LV as opposed to the mitral valve itself, and thus deformation of the closed valve toward the LV interior can result in an increase in measured volume. This measurement approach is consistent with the method of disks commonly used with MRI data (4, 27, 32) and various methods of volume determination used with echocardiography (26), all of which exclude the complex geometry of the LV at the leaflets and measure the LV chamber volume as it is defined by the extent of the myocardium.

Fig. 1.

Left ventricular (LV) volume is determined using disk summation from a short-axis stack of images. Disk edges are smoothly jointed to adjacent disks and combined with base and apex positions from long-axis images, allowing fractional inclusion of disks.

Imaging studies.

MRI provides a more reproducible measure of LV volume than echocardiography (4), is similar to echocardiography for determination of principal tissue strains (2, 3), and is similar or superior to echocardiography for the quantification of ventricular rotation or twist (14, 16, 23). MRI was used to measure the time course of LV volume because it is important for our study that LV volume be reliably measured without using geometric assumptions of LV shape employed by one- or two-dimensional echocardiography (Simpson's biplane or area-length methods). MRI was also used to measure the time courses of principal LV tissue strains (circumferential, longitudinal, and radial) and LV twist to allow for a direct comparison of these measures with LV volume; using MRI, strain and twist data were acquired within 10 min of volume data, without moving the subject, and under identical physiological conditions. However, MRI does not provide a robust assessment of mitral leaflet geometry or motion, and thus echocardiography was used to measure mitral leaflet motion throughout IVR.

MRI evaluation.

All MRI images were acquired using a 1.5 T scanner (Sonata; Siemens Healthcare, Erlangen, Germany). Image acquisitions were cardiac gated (electrocardiogram) and performed during breath holds at end expiration. Blood pressure was monitored using an automated blood pressure cuff.

Cardiac cine images of the two-, three-, and four-chamber long-axis views and a short-axis stack spanning the full LV were acquired using a steady state free precession (SSFP) sequence. Typical sequence parameters were as follows: 8 mm slice thickness, 2 mm gap between short-axis slices, 1.5 ms echo time (TE), 3.0 ms repetition time (TR), (94–146) × 256 matrix, 65° flip angle, (207–288) × 400 mm2 field of view (FOV), rate 2 parallel imaging, and 12 views per segment (VPS) for an acquired temporal resolution of 39 ms per cardiac phase.

Myocardial tissue tagging was used to measure tissue deformation. In the short-axis orientation, a gradient echo sequence with 8-mm grid tags was used for five evenly spaced slices spanning the length of the LV. Tags were applied following a delay of 200 ms from the electrocardiogram R wave to ensure the persistence of tags throughout diastole. Typical sequence parameters were as follows: 8 mm slice thickness, 8 mm gap between slices, 2.8 ms TE, 4.6 ms TR, 75 × 192 matrix, 14° flip angle, (200–225) × 400 mm2 FOV, and 5 VPS for an acquired temporal resolution of 23 ms per cardiac phase. In the long-axis slice orientations (two-, three-, and four-chamber), linear tags were spaced 8 mm apart and perpendicular to the long-axis of the LV with similar sequence parameters.

Phase-contrast MRI was used to acquire blood-velocity maps. The through-plane component of velocity was acquired for a short-axis slice located at the mitral valve. Typical sequence parameters were as follows: 8 mm slice thickness, 3.2 ms TE, 4.9 ms TR, 75 × 128 matrix, 30° flip angle, 1.2 m/s velocity encoding strength, (187–275) × 400 mm2 FOV, rate 2 parallel imaging, 3 VPS, and two phase encoding steps for an acquired temporal resolution of 29.5 ms per cardiac phase.

Echocardiography evaluation.

Echocardiography studies were completed within 3 h of MRI studies for each subject using similar supine subject positioning. For one subject, there was a 2-wk delay between MRI and echocardiography studies. Two-dimensional transthoracic ultrasound (Vivid i; GE Healthcare Clinical Systems, Wauwatosa, WI) was used to image the motion of the LV base and mitral valve during IVR using frame rates of 60–90 Hz. Cines in the apical four- and three-chamber and parasternal long-axis image orientations were acquired for each subject. Subjects were instructed to hold their breath at end expiration for all image acquisitions, similar to MRI studies.

Data analysis.

All data were analyzed offline (Matlab; MathWorks, Natick, MA). Aortic valve closure (AVC) and mitral valve opening (MVO) times were estimated using through-plane phase-contrast blood-velocity maps of a short-axis slice intersecting or near the mitral and aortic valve planes. Phase-contrast images were temporally interpolated to 10 ms/frame using piecewise cubic-spline interpolation. Blood-velocity data were used to estimate AVC and MVO times by extrapolating the velocity-time curves to their baseline or no-flow condition for the valve, as illustrated in Fig. 2. The baseline blood-velocity value at the aortic valve was assumed to be zero, reflecting no flow at the time of AVC, whereas a small nonzero baseline velocity was routinely indentified at the mitral valve.

Fig. 2.

The aortic flow velocity measured in the aortic valve (dashed curve) and the mitral flow velocity measured at the mitral valve (solid curve) are shown. The intersection of extrapolated aortic and mitral flow velocities and their respective baseline (no flow) values indicate aortic valve opening and mitral valve closure times, respectively. AVC, aortic valve closure; MVO, mitral valve opening.

LV volume was determined by the method of disks (disk summation) using the short-axis stack of SSFP images according to Simpson's rule (4, 27, 32). Endocardial and epicardial contours were manually traced on the short-axis stack, and papillary muscles and trabeculae were included as part of the LV lumen as shown in Fig. 1. The inclusion or exclusion of basal slices has been described as the major source of error in volumes measured using MRI (27). To address this issue, long-axis images were used to identify the base and apex of the LV (Fig. 1), allowing for the fractional inclusion of slices, similar to previously used techniques (18, 25). An identical method was used in a previous study to determine end-systolic and end-diastolic volumes and yielded interobserver and intraobserver variabilities based on the coefficient of variation of 2.6% and 2.1%, respectively (22). Early diastolic volume-time curves were generated by determining volume for all cardiac phases from AVC to the end of early filling.

Myocardial tissue deformation was assessed by tracking material points on the LV myocardium using tagged cines. An example of tracked points between two cardiac phases is shown in Fig. 3. Long-axis linear tags were manually tracked and used to assess longitudinal strain, whereas short-axis grid tags were tracked using an image-morphing technique and used to assess circumferential strain, radial strain, and untwisting. Analysis of short-axis grid tags using image morphing was performed using custom software similar to previously published methods (11, 20, 22). The spatial positions of each tracked point for both short-axis grid and long-axis linear tags were interpolated in time, using piecewise cubic-spline interpolation, to a uniform temporal resolution of 10 ms (from the true acquired temporal resolution of 23 ms).

Fig. 3.

Short-axis images with grid tags are shown for 1 slice from 1 of the participants. The images on the bottom show every second acquired image between 264 ms and 654 ms from the ECG R wave. The 2 enlarged images (cardiac phase 1 and cardiac phase 2) demonstrate the tracking of points over time. White circles indicate the position of tissue at cardiac phase 1, and gray circles indicate the position of the same tissue at cardiac phase 2. Arrows indicate the movement of the tissue calculated using deformation fields from image morphing.

Strain as a function of time was calculated using the change in length of a tissue segment at a given time relative to its reference length at diastasis, then divided by this reference length, in circumferential, longitudinal, and radial directions. Tissue lengths are defined as the distance between adjacent points in the circumferential and longitudinal directions and as the difference between distances from the LV centroid to adjacent points in the radial direction. For circumferential and longitudinal directions, both global and endocardial strains were measured; for the radial direction, only global strain was measured. Endocardial strains were calculated for the purpose of comparison with changes in LV volumes. The rate of change of strain was calculated as the discrete time derivative of the strain-time curve.

LV twist was calculated similarly to previous studies as the difference in rotations between the base and apex, taking into account the moving centroid of the LV mass at each cardiac phase (23, 33). A cardiac phase at diastasis was selected as the reference for the calculation of all rotation angles (10). The rate of untwist was calculated as the negative discrete time derivative of the twist-time curve.

The coefficient of variation representing interobserver and intraobserver reproducibility for the image morphing technique used to calculate circumferential strain, radial strain, and untwisting is shown in Table 1. Reproducibility is shown for the following key parameters: peak end-systolic circumferential and radial strains, peak diastolic circumferential and radial strain rates, peak end-systolic twist, and peak diastolic untwisting rate. The reproducibility of the timings of these key parameters is also shown.

View this table:
Table 1.

Reproducibility of strain and twist using grid tagging

To directly compare measured changes in strain during the isovolumic interval with corresponding changes in LV volume, the LV was modeled as half of an ellipsoid, symmetric about the LV long axis (26). The volume of a half ellipsoid is: Ve = 1/2[4/3πa2b] (Eq. 1), where a is the radius of the LV short axis at the base and b is the length of the LV long axis. To determine the initial dimensions of the ellipsoid, the end-systolic volume (ESV) and the eccentricity (γ, the ratio of the LV width to length) of the LV at end systole were used. ESV was directly measured using disk summation of short-axis images, and eccentricity was measured from four-chamber long-axis images. These values (ESV = Ve and γ = b/2a) were next substituted into Eq. 1, and a and b were solved for. Subsequent changes in ellipsoid dimensions were estimated from measured endocardial strains, where circumferential strain defines changes in the LV short-axis radius (a) and longitudinal strain defines changes in the LV long-axis length (b). With the use of Eq. 1, changes in these dimensions were used to estimate changes in LV volume associated with changes in circumferential and longitudinal endocardial strain. Endocardial strains were used because they correspond to the boundary of the LV cavity.

Long-axis cardiac ultrasound images were used to observe the changes in shape of the mitral valve leaflets during IVR. The deformation was measured in two ways using apical four- and three-chamber and parasternal long-axis views of the heart. First, the mitral valve leaflet was manually traced at the frame closest to AVC and immediately before MVO. Second, the shortest distance between the leaflet tips and a line intersecting the hinge points of the valve at AVC and MVO was measured to represent a quantitative measure of leaflet deformation.

Representation of data and statistical analysis.

All calculated parameters, including LV volume, twist, rate of untwist, strains, and strain rates, are represented as continuous time variables. These time courses are shown as representative curves produced by the average of the 10 subjects with SD indicated by error bars. The error bars are plotted at arbitrary, regular time intervals. When single values are quoted, they represent either peak-positive values, peak-negative values, or changes over specified time intervals. The timings of events were all normalized to AVC and expressed as either percentage of systolic duration or milliseconds from AVC.

Statistical analysis was performed using Microsoft Office Excel 2007 (Microsoft, Redmond, WA). Values are stated as means ± SD, and statistical significance was set at P < 0.05. Paired Student's t-tests were applied to test for differences in LV volume, strain, and twist measured between AVC and MVO times. Pearson's product-moment correlations and partial correlations, which control for time, were applied to represent the relationships between LV volume and other measured parameters (twist, circumferential, longitudinal, and radial global strains, and circumferential and longitudinal endocardial strains) during IVR depending on time and independent of time, respectively. To extract the correlational data, volume-time curves for each subject were referenced to their values at AVC and sampled every 10 ms, starting from 5 ms after AVC up to MVO, resulting in a series of discrete volume data. Corresponding data were also extracted for twist, global strains, and endocardial strains and were correlated with volume. Interobserver and intraobserver variabilities were calculated using the coefficient of variation.


Time courses of LV volume, global strain, and twist, and their respective rates of change, are plotted in Fig. 4. Peak values for all parameters and their timings along with the timings of other significant events are shown in Table 2. The observed sequence of events in all subjects was: 1) peak twist and strains (minimum circumferential and longitudinal strains and maximum radial strain) occurred at end systole, near AVC; 2) MVO occurred at 58 ± 10 ms after AVC (117 ± 3% of systole); 3) the peak rate of untwist occurred at 68 ± 25 ms after AVC (120 ± 8% of systole); and finally 4) the peak strain rates, circumferential and longitudinal stretching, and radial thinning occurred at 152 ± 21, 159 ± 19, and 138 ± 27 ms after AVC, respectively (145 ± 7%, 147 ± 7%, and 141 ± 9% of systole).

Fig. 4.

Representative time courses for volume change following aortic valve closure (A), twist and rate of untwist (B and C), circumferential (circ) and longitudinal (long) global strain and strain rate (D and E), and radial strain and strain rate (F and G) are plotted on the same time scale for comparison. Standard deviations are shown as error bars at arbitrary time points. The isovolumic relaxation (IVR) period is shown in gray.

View this table:
Table 2.

Peak values and rates of strain and twist throughout early diastole

Table 3 summarizes the changes measured in all parameters during the IVR interval. These IVR changes are also expressed as a percentage of the total change over early filling. LV volume underwent an apparent increase of 4.6 ± 1.5 ml, and circumferential, longitudinal, and radial strains changed significantly during IVR, with all changes in strain promoting an increase in volume. The averaged time courses for volume, twist, and strain rates over the IVR interval are shown in Fig. 5. The strain estimated volume change (3.0 ± 2.4 ml) using Eq. 1 was significantly smaller (P < 0.01) than the direct assessment of volume change, but the volume changes were significantly correlated (r = 0.83, P < 0.001), as shown in Fig. 6F. Similar comparisons were made between changes in measured LV volume and the changes in all strain components and with untwisting (Fig. 6). Table 4 summarizes the correlations between the apparent volume changes and other parameters, both as direct comparisons and also as partial correlations, controlling for time from AVC to correct for correlations due to similar time courses of volume and strain changes. The linear correlations between LV volume changes and changes in all other parameters were statistically significant (P < 0.01).

View this table:
Table 3.

Measured changes during isovolumic relaxation

Fig. 5.

Average time courses for volume change (A), twist (B), global and endocardial (endo) circumferential strain rate (C), global and endocardial longitudinal strain rate (D), and global radial strain rate (E) are plotted as a function of time from aortic valve closure. Standard deviations are shown as error bars at arbitrary time points. Mitral valve opening is indicated at time = 117% of systole.

Fig. 6.

Correlations between isovolumic changes in measured volume with global and endocardial circumferential strain (A and B), global and endocardial longitudinal strain (C and D), global radial strain (E), volume changes estimated from circumferential and longitudinal endocardial strains (F), and untwisting (G) are shown for multiple time points during isovolumic relaxation.

View this table:
Table 4.

Correlations with left ventricular volume increase during isovolumic relaxation

With the use of cardiac ultrasound, the leaflets were observed to deform toward the LV interior before their separation in all subjects, as shown in Fig. 7. The solid lines correspond to the leaflet position at AVC, and the dashed lines correspond to their position before MVO. A typical example for a parasternal long-axis view tracing is also shown in Fig. 7. The displacement of the mitral leaflet tips during IVR was 6.0 ± 3.3, 5.1 ± 2.4, and 2.1 ± 5.0 mm for apical four- and three-chamber and parasternal long-axis views, respectively. These displacements were not found to correlate with IVR changes in measured LV volume or components of principal myocardial strain.

Fig. 7.

Traced leaflet positions are shown at AVC (solid lines) and at MVO (dashed lines) for the apical 4-chamber and parasternal long-axis views for each of the 10 subjects. An example of the valve tracings is shown on the right for a parasternal view.


The major new findings of this study are: 1) MRI-derived measures of LV volume increase significantly during the IVR interval and 2) changes in measured LV volume over the IVR interval are significantly correlated with the changes in circumferential, longitudinal, and radial strain and with the extent of untwisting.

These findings support our hypothesis that imaging-derived volume increases measured during IVR are a consequence of myocardial expansion associated with increased fiber lengths during relaxation. The apparent increase in LV volume results from the measurement method, which excludes the complex boundaries defined by the closed mitral and aortic leaflets. Thus apparent LV volume changes can be viewed as representing regional changes in geometry, reflecting measurements being performed on the myocardium but not the valve leaflets. The measured myocardial expansion is thus likely balanced by the inward bowing of the mitral leaflets during this interval. The stiffness of the leaflets themselves has been shown to be actively modulated and to vary as a function of cardiac phase; stiffness is lower during IVR than during isovolumic contraction (19). Thus the change in measured volume may be modulated by the material and functional properties of the mitral leaflets in conjunction with the relaxation properties of the LV.

Isovolumic strain versus LV volume during IVR.

The measured volume increase during IVR, which neglects the inward deformation of the mitral leaflets, represents the overall expansion of the LV chamber and would be expected to correspond to the overall changes in principal tissue strains, lengthening in the circumferential and longitudinal directions and thinning in the radial direction. Overall, the changes in circumferential global and endocardial strain were best correlated with the volume changes (rcirc global = 0.86, rcirc endo = 0.86, P < 0.001). Longitudinal strain changes were less well correlated with the changes in volume (rlong global = 0.69, rlong endo = 0.63, P < 0.001), and the change in radial strain was least well correlated (rradial = −0.37, P < 0.01). The trends in these results are not surprising given their respective relationships with LV chamber volume. As indicated by the ellipsoid model used in this study, volume is linearly proportional to the long-axis length but is dependent on the square of the short-axis radius or, equivalently, the circumference (Eq. 1). Thus a given change in circumferential strain will give rise to approximately double the change in LV volume (for small changes in strain) compared with the same change in longitudinal strain. The changes in radial strain, unlike the circumferential or longitudinal strains, are indirectly related to LV volume changes due to its dependence on both the endocardial and epicardial borders. The volume-strain correlational data in Fig. 6 illustrate the relative strengths of these correlations. Changes in the amount of untwist were also highly correlated with volume changes (runtw = 0.81, P < 0.001). This relationship would not necessarily be expected because LV untwisting alone does not necessitate a volume change. However, when time from AVC was controlled for, this partial correlation coefficient was significantly reduced (runtw = 0.41, P < 0.01), whereas the partial correlation coefficient for circumferential strain was only slightly reduced with the same correction (rcirc endo = 0.75, P < 0.001). Thus the amount of covariation between measured changes in LV volume and untwisting may be attributed largely to their correlations with time; conversely, the covariation between measured changes in LV volume and changes in circumferential strain is largely independent of their correlations with time. Finally, volume changes estimated using the changes in circumferential and longitudinal endocardial strain using Eq. 1 yielded significantly lower volumes than direct measurement (3.0 ± 2.4 vs. 4.6 ± 1.5 ml; P < 0.01), but these were still significantly correlated (rvol est = 0.83, P < 0.001). This volume estimate is dominated by the circumferential strain component, as indicated in Eq. 1, and we would thus propose that global circumferential strain is the most practical surrogate for changes in LV volume representing the expansion of the myocardium during IVR.

Isovolumic changes in global circumferential, longitudinal, and radial strains have been measured using speckle tracking echocardiography (25, 36). Specifically, Wang et al. (36) measured a correlation between global longitudinal strain rate during IVR and the time constant of isovolumic LV pressure decline (τ). In their study, the global longitudinal strain rate was proposed to reflect the expansion of the myocardium associated with relaxation. Our current study confirms this supposition and further suggests that circumferential strain would be most representative of LV volume expansion associated with relaxation. Wang et al. (36) concluded that isovolumic longitudinal global strain rate reflects LV relaxation because the data were acquired during IVR, while the mitral valve is closed, isolating it from the effects of left atrial pressure. However, the current study illustrates that although the mitral leaflets are closed, they likely change their shape in conjunction with volume and strain changes. Thus the material properties of the leaflets and the atrial pressure would modulate the changes in strain during IVR in addition to LV relaxation.

Mitral leaflet motion during IVR.

The inward displacement of the leaflets toward the LV cavity was measurable in all subjects (Fig. 7) in all views and offers a source of blood displacement that can account for changes in tissue strain and apparent LV volume increases during IVR. This observed leaflet deformation is similar to previous studies of mitral valve function in isolated swine hearts (31) and in vivo ovine hearts (17). These studies concluded that the mitral leaflets descend into the LV throughout IVR, before the separation of the leaflets. However, the displacement of the valve leaflets in our study did not correlate with measured changes in volume. This may be due to the complex shape of the mitral leaflets; in other words, a one-dimensional measure of leaflet displacement may not be representative of the three-dimensional shape deformation. Additionally, it was not possible to quantify the volumetric displacement of the mitral valve leaflets during IVR using two-dimensional long-axis cardiac ultrasound images. However, a coarse estimation of the volume change can be made using the average displacements and annular radii. The base-to-apex displacement obtained by averaging the values measured in the different views is 0.43 cm. If a representative annular radius of 1.5 cm is assumed (estimated from Fig. 7) this results in a mitral valve area of ∼7 cm2. When this area is multiplied by the average displacement, a volume of ∼3 ml is obtained, similar to the volume change of 3.0 ± 2.4 ml predicted by the strain assessment. It is possible that three-dimensional echocardiographic imaging of the mitral leaflets at AVC and MVO could be used to provide a direct assessment of the apparent change in LV volume measured during isovolumic relaxation.


When results from this study are interpreted, it is important to consider the limitations of the imaging modalities used. Conventional cine MRI does not clearly resolve the aortic and mitral leaflets or their motion over time with sufficient resolution to identify AVC and MVO times. We approximated these times from blood-velocity-time curves measured near the aortic leaflet plane and at the mitral leaflet tips using the method illustrated in Fig. 2. With the use of this approach, the timing of key events such as peak twist (at 98 ± 5% of systole) and principal strains (minimum circumferential and longitudinal strains at 100 ± 5% and 102 ± 7%, respectively, and maximum radial strain at 98 ± 11%) were shown to be near the time of AVC in agreement with numerous other studies (11, 2225, 28, 33, 34, 36). The IVR duration (58 ± 10 ms in the current study) also is consistent with echocardiography guidelines for normal hearts in our subjects′ age range (21). The time of peak untwisting rate at 120 ± 8% of systole is also in agreement with literature value for healthy subjects (22, 24, 34). Finally, the percentage of untwisting at the time of MVO was measured to be 36.6%, which is in good agreement with previous echocardiography studies (11, 24, 28). Although all subjects did have a cardiac ultrasound evaluation, it was determined that the valve opening and closure times from these studies would not be accurate for use in analysis of the MRI studies due to potential differences in heart rate and loading conditions. Conversely, MRI evaluation of the valve timings occurred within 10 min of all other MRI data acquisition and under identical subject positioning and physiological conditions.

Although MRI has the advantage of being highly reproducible for the quantification of LV volume (4), it has the disadvantage of having lower temporal resolution compared with echocardiography. Thus a potential explanation for the larger volume changes measured from direct volumetric analysis (4.6 ± 1.5 ml), compared with the strain analysis (3.0 ± 2.4 ml), is the lower temporal resolution of the volumetric cine images, which could blur the larger changes in volume during early filling into the isovolumic period and lead to an overestimation of the apparent volume change. Nonetheless, the tissue tagging methods used for strain measures were acquired and analyzed independently of volume measurements with a higher temporal resolution and were significantly correlated with the volume data independently of their temporal correlations. This suggests that the apparent volume changes do represent the expansion of the LV myocardium during the isovolumic interval.

The interpretation of changes in strain, volume, or leaflet position over the IVR interval detailed in this study assumes that both aortic and mitral valves do not leak during this interval, meaning that volume contained between mitral and aortic valves is constant while valves are closed during IVR. Changes in any of the measured metrics with a leak present in either valve during this interval thus cannot be interpreted in the context of LV relaxation or mitral leaflet motion alone.


Although closed mitral and aortic valves ensure true isovolumic conditions during IVR, circumferential and longitudinal stretching and radial thinning during this interval contribute to a change in LV shape, which can be measured as an increase in apparent LV volume before MVO. Changes in global circumferential strain were the most representative of this apparent volume change. The observed inward motion (toward the apex) of the mitral leaflets allows for the conservation of volume. The leaflet motion and the conformational changes in the LV during isovolumic relaxation likely reflect active relaxation, LV pressure decline, and the material properties of mitral leaflets.


R. B. Thompson is an Alberta Heritage Foundation for Medical Research Scholar and acknowledges the financial support of the Natural Sciences and Engineering Research Council of Canada. M. J. Haykowsky has a career award from the Canadian Institutes of Health Research. J. Cheng-Baron, K. Chow, J. M. Scott, and B. T. Esch were supported by doctoral graduate scholarships from the Natural Sciences and Engineering Research Council of Canada and the Michael Smith Foundation for Health Research.


No conflicts of interest, financial or otherwise, are declared by the author(s).


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
View Abstract