Assessing the cause of MR requires visualization of the anterior and posterior mitral valve (MV) leaflets, the mitral annulus, the chordae, the anterolateral and posterolateromedial papillary muscle, and left ventricular (LV) wall motion abnormalities. Although echocardiography (transthoracic and especially 3D transesophageal) is an excellent imaging of the MV which CMR cannot match, CMR can help in identifying MV morphology and MR mechanisms. With a high signal-to-noise ratio and an excellent blood–myocardium contrast-to-noise ratio, cardiac balanced steady-state free precession (bSSFP) imaging is the method of choice for cine (motion) images. Retrospective cardiac gating, corresponding to the continuous acquisition of both ECG and image signals collected over several consecutive heartbeats and used after image reconstruction, should be used to allow optimized temporal and spatial resolution. In the setting of atrial fibrillation, however, prospective gating may result in better image quality. According to recent SCMR recommendations and a consensus on the use of CMR in MR (12, 13), the morphology and motion of the MV apparatus can be studied using

Leaflet motion abnormalities can be described as classically defined by echocardiography (prolapse, flail) (Figure 1). However, a flail leaflet in CMR can be masked by the regurgitant flow. Maximum mitral leaflet thickness is measured in diastole. Standard long-axis cine images are also informative for visualizing and sizing the mitral annulus (14). With CMR reference ranges, mitral annular dilatation, one of the mechanisms in secondary MR, can be identified. The mitral annular diameter is measured from the anterior commissure to the posterior commissure in the diastole (Figure 1). In primary MR (mitral valve prolapse especially), CMR can detect mitral annulus disjunction, a frequent component of myxomatous mitral valve disease (15) using standard long-axis cine images.

The LVOT-1 view can be used for determining the MV leaflet length, MV tenting, and tenting height, major determinants of ischemic MR. However, there are no data comparing CMR measurement of the mitral annulus or MV length to surgical findings, and echocardiography remains the gold standard in the perioperative assessment of primary MR.

On cine images, regurgitant jets can be visualized as a signal loss (dark/low signal) created by the turbulence of flow in the left atrium. The jet can be difficult to be clearly identified on fixed 2D images, especially if the direction of the jet is not linear. Moreover, the size and length of the jet can be reduced or increased by just changing parameters or sequences. Rapid spoiled gradient-recalled echo sequences with longer repetition and echo times or gradient echo (GRE) or hybrid GRE echo-planar sequences have been proposed for higher sensitivity (13, 17). The identification of such a flow void provides gross information about the location and direction of the jet, which can help in determining the etiology of the MR (for example, a central or eccentric jet, suggesting secondary or MV prolapse, respectively) (Figure 1). Compared to gradient-echo imaging, SSFP imaging results in significantly reduced signal voids in regurgitant jets in general and can easily underestimate the degree of regurgitation. These sequences are much less sensitive for identifying regurgitation than echocardiography and do not represent a reliable method for evaluating MR severity.

The most widely used CMR method to quantitatively assess the severity of MR is an indirect method combining 2D cine imaging and phase-contrast velocity mapping to quantify the regurgitant volume and fraction.

Mitral regurgitant volume (MRvol) is expressed as the difference between the left ventricular stroke volume (LVSV) and aortic forward flow (AFF), and the regurgitant fraction (RF) as the MRvol divided by the LVSV, expressed as a percentage.

The LVSV can be obtained from multiple LV short-axis cine bSSFP images, and the AFF from phase-contrast velocity flow in a trough-plane acquired at the level of the sinotubular junction or the aortic valve but not at the mid-ascending aorta due to the risk of overestimation (18). This sequence generates two types of images: a magnitude image showing the anatomy of the aortic valve and a phase map encoding the velocities within each voxel. A dedicated post-processing analyzes the blood flow through the aortic plane generating a flow curve, which allows the calculation of the aortic forward flow (AFF) (Figures 2, 3). An average of two to three flow measurement acquisitions is optimal.

An alternative and simple approach using only bSSFP sequences in the short axis is the “volumetric method,” expressed as the difference between LVSV and right ventricular stroke volume (RVSV), which represents the quantity of MR in the absence of other types of valvular regurgitation or intracardiac shunts.

Direct quantification of MR flow is theoretically feasible with phase-contrast velocity mapping by directly measuring the regurgitant flow across the MV.

However, given the significant dynamic motion of the MV during systole, high velocity, a non-circular orifice, and jet angulation, positioning a fixed slice across the MV (without annular valvular tracking) can be challenging and inappropriate, especially for patients with eccentric MR, multiple jets with different directions, and high heart-rate variability. Therefore, we do not recommend this approach for clinical practice.

4D flow velocity-encoded CMR imaging is an emerging technique that involves phase-contrast acquisition with flow encoding in all three spatial directions and to the dimension of time (3D + time = 4D). It allows the visualization of flow in multiple orientations to follow the spatial motion of the heart valves and the variable direction of the jets over time. The use of such a retrospective tracking method in a single volume acquisition covering the entire heart may enable direct MR jet volume quantification, overcoming certain limitations of 2D sequences (25). Recent reports have demonstrated its feasibility in mild to severe MR (26). However, its accuracy has usually been poor in comparison to 2D flow, especially in primary MR (25). This promising technique, which is constantly improving, is not yet commercially available for all scanners. Its principal disadvantage is the long scanning time (up to 10 min). Furthermore, it requires a post-processing software. Another limitation for direct measurement of MR severity is also the high velocity of the MR jet (> 5 m/s), which may result in aliasing and which can be more difficult to correct in 4D flow imaging and may require multiple velocity encodings.

Late gadolinium enhancement (LGE) imaging, acquired after gadolinium administration, has the unique ability to provide tissue characterization of the myocardium by assessing and quantifying regions of fibrosis and scars across the left atrium (27), the LV (28), and the papillary muscle (29).

Late gadolinium enhancement sequences [pulse sequences according to published guidelines (12)] should be performed at least 10 min after gadolinium injection in the short axis and the three long-axis planes (13). High-resolution dark blood LGE CMR is required to identify small papillary muscle enhancement (30).

In ischemic MR, the quantification of MR and myocardial infarct size with CMR can help for risk stratification in patients with advanced ischemic cardiomyopathy (31). In primary MR, especially MV prolapse, a high prevalence of focal replacement fibrosis, particularly in the segments adjacent to the posterior papillary muscle, can be observed (29). However, the LGE technique is not able to characterize diffuse fibrosis due to the absence of a normal myocardium as the reference. A recent parametric mapping technique (T1 mapping) allows for better characterization of global myocardial tissue composition by direct measurement of T1 relaxation times before and after contrast administration (32) and by calculation of the extracellular volume (ECV) (33).

In MR, little is known about diffuse interstitial fibrosis. One recent study suggested that in primary MR, ECV increases with MR severity, regardless of MV prolapse or the presence of replacement fibrosis (34). Diffuse interstitial fibrosis in primary MR may be more closely related to chronic LV volume overload than the etiology of MR.

Most studies have shown only limited agreement in the assessment of MR between CMR and echocardiography (20, 37–39). MRvol values truly differ among volumetric methods and the PISA method. There is a consistent tendency to obtain higher MRvol values by PISA relative to CMR (40–42), resulting in higher MR severity assessment with a risk of undergoing surgery without severe MR by CMR (20). The discrepancy between these two modalities is more pronounced in patients with late systolic or multiple jets than central and holosystolic jet (38). Therefore, using the same thresholds to define MR severity and the need for surgery may be inappropriate. The present available literature indicates that PISA MRvol thresholds and MRvol values by CMR are different and not interchangeable. Interestingly, the MRvol was similar using the volumetric method by 3D echocardiography and CMR (43). Recent reports have suggested that 3D-transthoracic echocardiography is less dependent on geometric assumptions and allows better visualization of the vena contracta regurgitant orifice. However, characterization of the Doppler jet is still challenging, with additional limitations of lower spatial and temporal resolution (36).

There is a paucity of data on specific CMR thresholds to define MR severity due to the lack of large trials with a validation cohort. Gelfand et al. (44) proposed that RF CMR cutoff value of 42% (indirect method) for severe MR correlates well with Doppler echocardiography. Interestingly, one study used the multiparametric approach by Doppler echocardiography as the reference standard and compared it to CMR RF. MR severity graded by two experts, showed excellent agreement. The authors observed that significant MR (moderate to severe or severe) could be very accurately identified by CMR using a RF cutoff value of 35% (45).

Myerson et al. (46), on initially asymptomatic patients with moderate or severe MR followed up for up to 8 years, showed that MRvol and RF by CMR were the most discriminatory parameters to determine the need for surgery (symptoms or other indications for surgery), with cut-off thresholds of 55 ml for MRvol and 40% for RF, and an increasing risk with increasing values of the parameters. The RF cutoff for severe MR in this study was very similar to a recent study by Polte et al. (47) (RF > 41%) and lower than echocardiographic criteria (RF ≥ 50%). As the MRvol is proportional to LV size in primary MR, for two patients with similar CMR MRvols, the patient with a smaller LV will have more severe MR; calculation of the RF may overcome this issue as it normalizes the MRvol to the size of the LV.

Interestingly, integrating ECV may also provide an additional benefit for the selection of asymptomatic patients for mitral correction. Recently, Kikungvan et al. (48), in a prospective observational registry of patients with at least moderate primary MR, showed that RF and elevated ECV were independently associated with events, with a cutoff of 40% for RF and 30% for ECV for mitral surgery.

Another approach to evaluate the accuracy of an imaging modality to quantify MR is to assess LV remodeling after surgery. Some authors have found a good correlation between MRvol by CMR and a decrease in LV volume post-MR correction (20, 40).

With a high contrast between the valve leaflets and blood pool, together with a high signal-to-noise ratio, SSFP imaging can be used to visualize aortic valve anatomy in any plane, irrespective of cardiac anatomy (50). A perpendicular plane through the coronal oblique LVOT view allows the visualization of each cusp (left, right, and non-coronary). Contiguous cine images 3 mm above and below this slice are required (51). During diastole, three normal leaflets form a three-pointed star “Mercedes-Benz emblem,” but a bicuspid aortic valve can be identified (Figure 4).

Assessing the aortic diameter is crucial for determining AR etiology and follow-up. The diameter of the ascending aorta can be measured using non-contrast-enhanced MR angiography (NCE-MRA) with a respiratory navigator or breath-hold contrast-enhanced MRA (CE-MRA). Measurements are performed at end-diastole using the internal diameter of the aorta at the level of the sinus, the sinotubular junction, and the level of the pulmonary trunk on the ascending aorta (52) or at the level of maximum dilatation. Ideally, measurement should be performed in 3D using a double oblique angulation perpendicular to the vessel axis. Measurement in standard 2D axial view should be avoided.

Aortic regurgitant jets can be visualized on cine images using long-axis LVOT views (3-chamber and coronal oblique LVOT) as a dark jet due to a signal void from turbulence back flowing into the LVOT during diastole (Figure 4). The location and direction of the jet can provide additional information about the AR etiology. For example, a central jet can be more highly related to aortic dilatation and an eccentric jet more to a cusp abnormality (prolapse) (Figure 4). However, the appearance of the signal void is dependent on the hemodynamics, echo time, and flip angle (53), and does not directly reflect the aortic regurgitant volume (ARvol); therefore, a more reliable assessment is required.

Cardiovascular magnetic resonance has the ability to measure flow velocity and direction over time. Unlike MR (with multiple, turbulent, irregularly shaped jets), direct flow quantification of AR is feasible. The most commonly used CMR method to quantitatively assess the severity of AR is a direct measurement using through-plane velocity mapping performed just above the aortic valve (2D phase-contrast imaging: 2D-PC) in a plane perpendicular to the direction of blood flow (21). A dedicated post-processing analyzes the blood flow through the aortic plane generating a flow curve, which allows calculation of the AFF, aortic regurgitant volume (ARvol = area under the backward flow curve during the diastolic phase of the cardiac cycle), and RF (ARvol/AFF) (54) (Figures 3, 5).

ARvol can be measured indirectly as the difference between the AFF and PFF in the absence of other regurgitation or intracardiac shunts.

Similar to MR, the volumetric method can be used to quantify ARvol using only bSSFP sequences in the short axis. ARvol is expressed as the difference between LVSV and RVSV in the absence of other types of valvular regurgitation (54).

4D flow with its ability for measuring eccentric, non-laminar flow in any orientation in space could be in the near future appropriate for AR quantification (59).

T1 mapping, with its potential to assess cellular and extracellular compartments, could be of interest in AR as a prognostic marker for clinical outcomes (60).

Unlike MR, there is a paucity of large and prospective studies comparing these two modalities. In all studies, the direct CMR method was used and thresholds to define severe AR were the same as those used for echocardiography: Rvol ≥ 60 ml and RF ≥ 50% according to recommendations (35, 36, 61) (Table 3). In most studies, the correlation between these two modalities was modest (62–65), particularly in eccentric jets (66).

2D PISA is limited to the alignment of the flow and by the need for computation and geometric assumptions. 3D echocardiography, not restricted by any imaging plane, could overcome these limitations and provide a better correlation with CMR, especially in eccentric (67).

As previously noted, ARvol and/or RF measured by CMR are categorized based on established echocardiographic guidelines (35), defined as:

As for MR, there is a paucity of data on specific CMR thresholds for ARvol and RF, with a lack of large trials with a validation cohort. ARvol > 40 ml and RF ≥ 30% have been proposed to define severe AR with CMR that correlates best with echocardiography (47, 62) or for identifying patients who developed symptoms or needed aortic correction surgery (68). These CMR thresholds are much lower than established echocardiographic guideline criteria (Rvol ≥ 60 ml and RF ≥ 50%) (35, 36) and could explain the frequent mismatch between moderate to severe AR by 2D echocardiography and mild to moderate AR by CMR (65).

In the absence of a reference method, it is difficult to determine which imaging modality can better predict the need for aortic correction, particularly for asymptomatic patients. However, several studies have reported a better prediction with CMR (64, 65, 69).

Echocardiography is the first-choice tool to grade AR based on qualitative, semiqualitative, and quantitative criteria (4). Current class I surgical recommendations for severe AR are based on symptoms or for asymptomatic patients with LV dilatation or dysfunction (LVESD ≥ 50 mm or > 25 ml/m² or LVEF ≤ 50%). CMR is a “second tool,” indicated when various echocardiographic parameters are inconsistent (Table 4). Although there is a lack of large, prospective, comparative studies, the publications discussed above show that CMR could improve the diagnosis and surgical timing of patients with AR relative to echocardiography (Tables 5, 6). CMR should be considered for patients with severe AR by echocardiography to confirm the severity and guide surgical decision-making. From a practical point of view and in light of studies discussed above, AR is very likely to be significant if aortic RF ≥ 30%, even if large trials to define specific CMR thresholds are needed.

Due to the complex interplay between TR and the RV (TR can cause RV dilatation and vice versa), assessing TR requires the accurate assessment of RV structure and function. CMR is the method of reference to assess RV volume, function, and mass, without geometric assumptions (74).

RV structure can be studied using

The RV regional wall-motion abnormalities can be determined visually. Recently, feature tracking, similar to speckle tracking in echocardiography, has emerged to more accurately detect changes in wall motion and earlier ventricular dysfunction, which may be an independent predictor of mortality in severe TR (76).

Tricuspid anatomy can be assessed by short- and long-axis views, as already described, allowing simultaneous visualization of the three leaflets (septal, anterior, and posterior) in the short-axis view. Leaflets can be described as prolapsed, restricted, thickened, or with tenting (77). Tenting height and area may be measured and considered abnormal if they are > 7 mm (3 mm/m²) and 1.1 cm² (> 0.5 cm²/m²), respectively (14). However, tricuspid leaflets are thin, as well as the chordae and papillary muscles, and can be difficult to visualize. Therefore, echocardiography remains the cornerstone for TR anatomy assessment.

The tricuspid annular diameter, a relevant parameter included in the guidelines for tricuspid intervention, can be measured in the 4-chamber view in early diastole, with the upper limits of the normal value recently reported to be 43 mm (22 mm/m²) (78) (Figure 6). It is important to note that for patients with pacemakers or defibrillator leads, generated metallic artifacts decrease the accuracy of CMR for the assessment of RV volume structure and tricuspid anatomy, as well as for patients with arrhythmias.

The qualitative assessment of TR can be performed using bSSPF imaging (2 specialized long-axis and 4-chamber views). TR is visualized as a dark jet created by flow turbulence generating spin dephasing back flowing into the right atrium. As for MR and AR, this is a gross assessment and is not reliable on its own. In the presence of a turbulent jet, regurgitant flow appears as a low signal that can mask the visualization of the leaflets.

Similar to MR, the regurgitant volume of TR (TRvol) can be measured indirectly as the difference between RVSV and PFF (or AFF in the absence of AR) using 2D cine (for RVSV) and retrospective acquisition with velocity-encoded phase-contrast sequence in a through–plane on the pulmonary valve (or the aortic valve), 5 mm above the valve (12) (Figure 7).

This is the most widely used CMR method and is valid in the absence of intra-ventricular shunts (79). The advantages and technical considerations are similar to those for MR and have already been discussed in the MR chapter.

As for MR and AR, the volumetric method can be used to quantify the TRvol, expressed as the difference between RVSV and LVSV in the absence of other forms of valvular regurgitation. The advantages and technical considerations are similar to those for left-sided valvular regurgitation and have already been discussed in the MR and AR chapter.

These two indirect methods can be performed in addition to echocardiography to increase the confidence in the assessment of TR severity.

2D PC imaging also allows the direct measurement of TR regurgitant flow through the tricuspid valve. As for MR, this technique is limited by significant motion and the saddle-shape of the tricuspid annulus and is not commonly used in clinical practice.

Data concerning 4D flow in TR are very scarce. One recent study have suggested high concordance of 4D flow with standard 2D PC (indirect method) (80).

Unlike LV, tissue characterization (LGE and T1 mapping) of the RV is more complex due to the thin RV wall (thickness 1–3 mm), which creates partial volume effects. Transmural RV enhancement can be visualized, but non-transmural lesions are very challenging to visualize. LGE patterns may help to understand the underlying mechanism of TR, as it allows to detect ischemic patterns associated with pulmonary hypertension, or fibrosis associated with cardiomyopathies such as arrhythmogenic right ventricular cardiomyopathy. T1 assessment in TR has not yet been reported.

As for MR and AR, using echocardiographic thresholds is inaccurate and extrapolating data from MR is too simplistic given the differences in anatomy, regurgitant orifice and shape, and hemodynamics between the two valves. A recent study by Zhan et al. proposed a different approach by defining optimal CMR thresholds for TRvol and RF for a low, intermediate, and high risk of mortality (81). The study by Zhan et al. has the merit of providing CMR thresholds for mortality risk, independent of the strong confounding effect of RV dilatation and dysfunction. Due to the lack of randomized studies using these CMR thresholds, the benefit of TR correction (surgery or percutaneous device) is still unknown. Current guidelines, mostly based on expert opinion, recommend intervention for patients with severe TR and concomitant left-sided valve surgery (class I) (4). In the setting of severe isolated TR, surgery is recommended for symptomatic patients without RV dysfunction (class I) or asymptomatic patients with RV dilatation or RV dysfunction (class IIa), without defined thresholds. However, CMR could contribute to defining these thresholds. Indeed, in a cohort of 76 patients undergoing tricuspid surgery, Park et al. found that CMR-based RVEF was an independent predictor for cardiac mortality and major post-operative cardiac events, with a cutoff of 46% (82). A recent observational prospective study by Rodriguez et al., on 43 patients with severe echocardiography-based TR undergoing tricuspid annuloplasty suggested that RVEDV indexed by CMR is the best independent predictor of overall mortality at follow-up, with a cut-off of 104 ml/m², associated with higher cardiovascular mortality (83).

Aside from further pending CMR studies on specific thresholds, those defined by Zhan et al. (81) could be used in clinical practice and added to the established relevant value of CMR for RV volume and function.

The assessment of RV structure and function is essential for determining the prognosis and planning surgical/percutaneous intervention for patients with rTOF (84).

Contrary to echocardiography, CMR provides good visualization of RVOT anatomy using a stack of SSFP cine images in the short and long axes (RVOT and 3-chamber). The anatomy of the pulmonary valve, with three semi lunar leaflets (anterior, left and right), can be visualized in through-plane images on the valve (Figure 8).

PR can be seen as a dark jet in cases of flow turbulence in diastole, extending into the RVOT. However, in cases of laminar flow with a wide jet, there is almost no turbulence, making the jet nearly invisible. As for other forms of valvular regurgitation, it is a gross assessment and is not alone reliable.

Similar to AR, direct measurement of pulmonary backward flow is feasible using 2D-PC with retrospective acquisition in a through-plane acquisition on the pulmonary valve, 5 mm to 1 cm above the level of the valve (12) (Figure 9).

This is the most widely used method and the current reference standard technique (85). In patients with rTOF or dilated pulmonary artery, pulmonary flow can be non-laminar and turbulent, leading to a possible underestimation of pulmonary regurgitant volume (PRvol). In this case, the volumetric method or 2D-PC acquisition perpendicular to the right and left pulmonary arteries can be performed to confirm the findings.

As for other forms of valvular regurgitation, the volumetric method can be used to quantify PRvol by comparing RVSV and LVSV on bSSFP short-axis cine images (Figure 9), in the absence of other regurgitant disease or intra-cardiac shunt.

PRvol can be measured indirectly as the difference between PFF and AFF in the absence of tricuspid regurgitation (Figure 9).

PRvol can be measured indirectly as the difference between RVSV and AFF in the absence of tricuspid regurgitation.

The presence of focal RV and LV enhancement in patients with rTOF is a common finding, mostly located at surgical sites of the RVOT, the patching of the ventricular defect, and in the inferior RV insertion (86). Such focal fibrosis can be a substrate for arrhythmias. The RV ECV could be associated with major adverse cardiovascular events (87). However, due to the thin walls of the RV, the proximity to the blood pool, and epicardial fat, RV ECV measurement is subject to potential errors.

The CMR thresholds to define significant PR are the subject of debate. Several published studies have classified PR as significant if the RF ≥ 20% (91, 92) and others have used three grades of RF: mild < 20%, moderate: 20–40%, and severe > 40% (88, 89). Surprisingly, specific CMR PRvol thresholds have not been yet defined, despite a study showing better identification of severe RV dilatation using the PRvol than RF (92). Severe PR results in significant RV dilatation and eventually RV dysfunction, which can be irreversible, particularly for patients with rTOF. In this population, many centers recommend early PV intervention before symptoms or marked RV dilatation. CMR plays a crucial role in defining the optimal timing for PR intervention due to its ability to measure EF and RV dilatation with high accuracy. Geva et al. proposed a surgical approach for patients with a RF ≥ 25% and at least two of the following parameters: indexed RV end diastolic volume ≥ 160 ml/m², indexed RV end systolic volume ≥ 70 ml/m², indexed LV end diastolic volume ≤ 65 ml/m² and RVEF ≤ 45% (84). Since the development of percutaneous valve procedures as an alternative to surgery, lower thresholds may be applied, although studies are lacking to provide guidance.

As for the other regurgitation, CMR is indicated in the case of inconclusive echocardiography (Table 4); however, CMR is the preferred method of non-invasive imaging to quantify PR, follow patients with rTOF, and guide the surgical intervention.