Key words
heart - MR-imaging - cardiac function - myocardial relaxation time mapping - myocardial
perfusion - pitfalls
Introduction
Currently, cardiac magnetic resonance imaging (MRI) represents the reference standard
method in many clinical situations both for the evaluation of heart function as well
as for non-invasive tissue analysis of the myocardium [1]
[2]
[3]. The quantification of cardiac and myocardial, functional and morphological parameters
plays an increasingly central role in general cardiac diagnostics and differential
diagnosis. While volumetric and blood flow-based cardiac MRI parameters can be used
to define ventricular and atrial function, myocardial hypertrophy and dysfunction,
as well as shunt and heart valve regurgitation volumes [4]
[5], relaxation time mapping techniques allow quantitative morphological differentiation
of global and regional myocardial changes, e. g. in the context of myocarditis, cardiac
amyloidosis, Anderson-Fabry disease or cardiac iron storage disorders [6]. In addition, the quantification of myocardial perfusion, which could contribute
to the diagnosis of multi-vascular and non-obstructive coronary vascular diseases,
is becoming increasingly employed [7]
[8].
This review aims to summarize established and new, quantitative routine cardiac MRI
parameters as well as describe their significance and interconnections. A major challenge
in the interpretation of quantitative cardiac MRI parameters lies in the definition
of their normal values, which typically depend on both the scanning technique and
evaluation algorithm [9]. Although the technical basics of cardiac MRI = Magnetic Resonance Imaging are only
briefly discussed and reference is made to detailed reviews (among others [10]
[11]
[12]
[13]
[14]), substantial influencing factors on quantitative cardiac MRI parameters will be
presented in more detail.
Cine Imaging
Cine imaging refers to the time-resolved representation of individual cardiac phases
during a heartbeat. The k-space-segmented balanced steady state free precession (bSSFP)
sequence is currently the standard technique for acquiring cine series. For patients
with metallic cardiac implants (heart valves, pacemakers, implantable cardioverter
defibrillators, etc.), the k-space segmented flash sequence (fast low-angle shot)
is an alternative imaging technique with reduced artifact susceptibility [13]
[15]. Cine imaging during irregular heartbeat and/or respiration can be performed using
real-time cine protocols. Using new compressed-sensing algorithms, temporal resolutions
(≤ 45 ms [1]) and spatial resolutions comparable to k-space-segmented standard sequences can
be achieved [16].
Volumetric Function Parameters
Cardiac MRI is considered the reference standard method for determining volumetric
ventricular function parameters. In the clinical routine, standard evaluations include
the left ventricular (and, depending on the clinical question, the right ventricular)
end-diastolic (EDV) and end-systolic (ESV) volumes, stroke volume (SV), cardiac output
(CO), ejection fraction (EF) and myocardial muscle mass [1]
[2]. Volumetric quantities are given as absolute values or normalized to the body size.
Due to their high reproducibility and accuracy, these are important parameters in
cardiac diagnostics as well as for therapy and follow-up controls. It should be noted
that volumetric cardiac MRI function parameters differ methodologically from echocardiographic,
computed tomography or single-photon emission computed tomography (SPECT) function
parameters limiting comparability of parameters throughout techniques [17]
[18]
[19].
Apart from the dependence of volumetric cardiac MRI function parameters on gender
and age, both the acquisition technique (bSSFP or flash) and the evaluation strategy
(manual, semi-automatic or automatic segmentation, inclusion or exclusion of papillary
muscle, type of base level definition) must be taken into account when interpreting
the data and comparing it with normal values ([Fig. 1]). According to the guidelines for standardized cardiac MRI evaluation, it is recommended
to segment myocardial trabeculae and papillary muscles to the myocardium to quantify
muscle mass, but to assign them to the blood volume for the assessment of ejection
fraction and ventricular stroke volume [4]
[9].
Fig. 1 Impact of acquisition and evaluation on left ventricular volumetric function parameter
and myocardial mass. a bSSFP compared to flash-acquisition. b Trabeculae and papillary muscles included to the myocardium (Trab+) compared to trabeculae
and papillary muscles excluded from the myocardium (Trab–).
With the implementation of robust automatic segmentation algorithms, all cardiac phases
can be routinely evaluated volumetrically, and new functional parameters can be derived
from the ventricular volume-time curve and its time derivative (dV/dt) ([Fig. 2]); dV/dt curves represent ventricular blood flow rates in which the systolic minimum
represents the peak ejection rate (PER), the diastolic maximums the early peak filling
rate (PFRE) and the peak filling rate at atrial contraction (PFRA). The maximum filling rates, their ratios (PFRE/PFRA), and filling rates normalized to the end-diastolic volume (PFRE/EDV, PFRA/EDV) have the potential to characterize diastolic ventricular function and classify
diastolic dysfunction without additional measurements [20]
[21]. While ventricular function parameters are usually evaluated from cine short axis
image stacks covering the ventricles, additional cine series in short or long axis
orientation are required for volumetric evaluation of the left and right atrial function
([Fig. 2]). Based on the biphasic atrial volume-time curve, maximum and minimum atrial volumes
can be evaluated, and the total (TEF), passive (PEF) and contractile (CEF) atrial
ejection fraction can be calculated [22]. Atrial function parameters can – analogous to echocardiography – also be estimated
from biplanar area-length measurements, whereby atrial volumes and ejection fractions
are overestimated compared to volumetric evaluation [9].
Fig. 2 Schematic diagram of ventricular and atrial volume-time curves. a Ventricular volume-time curve (red graph) is shown together with its time derivative
dV/dt (blue graph), the maximal ventricular emptying (PER) and filling rates (PFRE, PFRA). b Atrial volume-time curve for determination of the total, passive, and contractile
atrial ejection fractions derived from maximal, minimal and before atrial contraction
volumes. ED, ventricular end-diastole; ES, ventricular end-systole.
Irregular heartbeat poses a methodical limitation of all volumetric cardiac MRI function
parameters. Although real-time imaging independent of respiration or heart rhythm
allows the acquisition of cine series with virtually no movement artifacts, in the
case of an irregular heartbeat, ventricular volumes differ generically from heartbeat
to heartbeat. In studies including subjects with irregular heartbeat, evaluation of
volumetric function parameters was performed by acquiring real-time cine series over
several heartbeats and selecting comparable heartbeats for analysis [23]
[24].
Myocardial Function Parameters
Together with the myocardial mass, myocardial wall thickness can be obtained from
the segmentation of the myocardium for the diagnosis of myocardial hypertrophy and
vitality, as well as systolic-to-diastolic wall thickness changes for the assessment
of global and regional myocardial kinetics [4]. Regional parameters are typically reported according to the American Heart Association
(AHA) 17-segment model. When evaluating the wall thickness, care must be taken that
the papillary muscle and trabecula are not segmented to the myocardium and that the
aortic outflow tract does not falsify the wall thickness of the basal anterior septal
myocardial segment.
Cine tagging imaging is considered the reference standard method for the analysis
of ventricular wall deformations (longitudinal, radial and circumferential strains)
and deformation rates (strain rates) [25]. Feature tracking allows the evaluation of ventricular and atrial strains as well
as strain rates based on routine cine series, allowing myocardial wall deformation
to be detected without additional measurements in routine imaging ([Fig. 3]). Strains and strain rates vary, however, both regionally (basal/midmyocardial/apical,
endocardial/epicardial) and depending on the acquisition technique (feature tracking,
displacement encoding with stimulated echoes (DENSE) imaging, cine tagging). In contrast
to the volumetric function parameters, strains and strain rates differ depending on
the specific evaluation software, which significantly limits the interpretation of
myocardial strains in clinical routine and their comparison with published normal
values [26]
[27].
Fig. 3 Ventricular strains and strain rates. a Schematic drawing illustrating the definition of circumferential, radial and longitudinal
strains. b In tagging-analysis, myocardial strains are derived from tracing the deformation
of an initially tagged grid (yellow lines). c In feature-tracking displacement of myocardial pixels during the heartbeat are modelled
from standard cine series. d Representative time courses of global left ventricular strains (red line graphs)
and their time derivatives (strain rates, blue line graphs).
Heart Valve Function
In cine series, jets caused by heart valve stenoses or insufficiencies are visualized
as signal cancellations; their appearance depends significantly on the parameters
of the acquisition sequence (especially echo time) and the turbulent nature of the
blood flow in the jet. Consequently, this phenomenon cannot be used to quantify cardiac
valve stenoses or insufficiencies. However, planimetric evaluation of cine series
acquired in the valve plane allow for determination of valvular opening area, and
valve stenoses can be graded in accordance with echocardiography, the reference standard
method [28]
[29].
Phase-contrast Imaging
The phase contrast technique allows the quantification of blood flow and myocardial
velocities in any spatial direction [30]
[31]. In cardiac MRI, the technique is typically used as a cine technique with unidirectional
velocity coding perpendicular to the acquisition plane (through-plane) and called
2D flow, whereby the two dimensions refer to time resolution and unidirectional velocity
measurement. Using acceleration techniques, 4D flow measurements, i. e. the recording
of the temporally resolved tridirectional velocity field in a volume, become applicable
for clinical routine [32].
2D flow measurements are typically acquired orthogonally to the assumed main flow
direction through a selected cross-section (e. g. orthogonally to a vessel or parallel
to atrio-ventricular heart valve planes). These can be used to determine temporal
changes of the cross-sectional area, the maximum velocity, average speed and blood
flow through the cross-section, as well as the corresponding temporally integral flow
volume in the cardiac interval. Using multiplanar reconstruction, analogous quantities
can be determined from 4D flow measurements for each cross-section within the recorded
volume ([Fig. 4]). Both methods offer different advantages and disadvantages. Advantages of 2D over 4D
flow measurement include short scan times (so that data can be acquired under breath
holding) and the possibility of optimize velocity coding (VENC, as small as possible
to maximize velocity-to-noise ratio, but greater than maximum velocities to avoid
aliasing and thus potential errors in evaluation [30]
[31]
[33]). Compared to 2D flow measurements, 4D flow measurements are characterized by simple
planning and a-posteriori analysis of any measurement plane of interest from a data
set [32].
Fig. 4 2D and 4D phase contrast imaging derived parameters in the aorta ascendens. a Segmentation of the aortic vessel cross-section (yellow line) in a 2D-flow measurement
with the imaging plane orientated orthogonal to the vessel course. b Multiplanar reconstruction of an evaluation plane (red plane) in a 4D flow measurement
covering the entire aorta ascendens for a posteriori evaluation of aortic blood flow.
Visualization of blood velocity field through the aorta (vectors) enables optimized
alignment of the evaluation plane orthogonal to the direction of the blood flow. c Representative time courses of the vessel cross-section area, average and maximum
through-plane velocity through the segmented vessel cross-section, and the calculated
blood flow through the vessel cross-section.
Quantification of Blood Flow Volumes
The central application of the phase contrast method in routine cardiac MRI is the
quantification of aortic (QA) and pulmonary blood flow volumes (QP) in the heartbeat. In the absence of cardiac/cardiovascular shunts and/or heart valve
insufficiencies, QA and QP correspond to the left and right ventricular stroke volume [5]
[34]. By comparing the blood flow volumes with each other (QP/QA ratio) and with the volumetric ventricular stroke volumes, both shunt volumes and
regurgitation volumes of the atrio-ventricular valves can be estimated ([Fig. 5]). Regurgitation volumes of semilunar valves can be determined directly from the
time course of aortic and pulmonary phase contrast measurements. While no effective
blood flow through the vessel cross-section can be detected after end-systolic valve
closure, valve insufficiencies cause a diastolic reflux volume ([Fig. 6]).
Fig. 5 Interpretation schemes for combined analysis of volumetric ventricular stroke volumes
and phase contrast imaging derived blood flow volumes a in absence of atrio-ventricular valve insufficiencies and absence of cardiac/cardiovascular
shunts, b in presence of atrio-ventricular valve insufficiencies and absence of cardiac/cardiovascular
shunts and, c in presence of cardiac/cardiovascular shunts. d Summary on relations between volumetric ventricular stroke volumes and phase contrast
imaging derived blood flow volumes. LA/RA, left/right atrium; PV, pulmonary vein;
VC, venae cavae; MINS/TRINS, mitral/tricuspidal insuffiency; PVF, anomalous pulmonary
venous connection; ASD, atrial septal defect; VSD, ventricular septal defect; PDA,
patent ductus arteriosus; LVSV/RVSV, left/right ventricular stroke volume; QA, aortic flow volume per heartbeat; QP, pulmonary flow volume per heartbeat, QP,R/QP,L, flow volume through the right/left pulmonary artery branch.
Fig. 6 Evaluation of the regurgitation volume in aortic valve insufficiency. a Representative time course of the blood volume through the aortic vessel cross-section
in absence of an aortic valve insufficiency. The forward blood volume after systolic
aortic valve closure equals the aortic flow volume within the cardiac cycle QA. b Representative time course of the blood volume through the aortic vessel cross-section
in presence of an aortic valve insufficiency. After systolic aortic valve closure
backward flow is observed through the aortic vessel cross-section. The regurgitation
volume is calculated as the difference between forward volume and QA.
The comparative interpretation of volumetric and phase-contrast-based blood flow volumes
is decisively determined by the accuracy of the volumetric evaluation, accuracy of
the phase contrast measurement and physiological variations between the measurements.
Volumetric ventricular stroke volumes can be verified using the (diastolic) inflow
volume measured through the atrio-ventricular valves [35]. The accuracy of the results of phase contrast measurements is limited by the segmentation,
the choice of measuring plane, as well as possible spatially variable background phases
(exact orthogonality to the vessel plays a secondary role, since the resulting underestimation
of the velocity is compensated by the overestimation of the cross-sectional area).
Background phases can be minimized by measuring near the isocenter and by adequate
correction in post-processing [30]
[31]. 4D flow measurements have the potential to further improve the accuracy of phase
contrast measurements, since physiological variations between individual 2D phase
contrast measurements are eliminated and (valve) planes can be adapted accordingly
over the course of time [32].
Maximum and Peak Velocities
Measurement of maximum and peak velocities plays an important role in the determination
of the severity of valve stenoses as well as in the evaluation of diastolic heart
function.
Peak velocities in jets caused by valve stenoses can be determined in accordance with
the reference standard method, echocardiography. The prerequisite for this, however,
is that in the phase contrast measurement short echo times and adequate temporal and
spatial resolution are selected, the 2D phase contrast plane in the vena contracta
is aligned orthogonally to the velocity, and velocities are evaluated without averaging
from individual pixels [4]
[30]
[31]
[36].
Analogous to echocardiography, the early (E) and late diastolic (A) peak blood flow
profiles over the atrio-ventricular valves can be used to determine the transmitral
or transtricuspidal E/A ratio as a parameter of ventricular diastolic function ([Fig. 7]). Other phase contrast parameters of the diastolic function are [37]
[38]
[39]: the systolic (S) and early diastolic (D) pulmonary vein peak velocity, as well
as the S/D ratio for grading diastolic dysfunction, myocardial early diastolic velocity
(e'), and E/e' ratio for estimating left ventricular filling pressure. Although studies
show a good correlation of these parameters to echocardiography [39], diastolic function parameters are routinely not evaluated due to the necessity
of additional measurements and the lack of standardization of acquisition protocols.
It should be noted that phase contrast protocols with higher temporal resolution (15–20 ms
[37]
[39]) and adapted VENCs (100–180 cm/s for transmitral, 10–30 cm/s for myocardial measurements
[37]) have been used in studies to evaluate diastolic functional parameters. With the
increasing implementation of 4D flow measurements, evaluation of diastolic function
without additional measurements could become established in clinical routine [32].
Fig. 7 Evaluation of diastolic left ventricular function parameters from transmitral 2D-flow
measurements. a Representative time course of the transmitral maximum velocity (red graph) with early
(E) and late (A) diastolic peak velocities. b Representative time course of myocardial tissue maximum velocity (red graph) with
the early diastolic peak velocity e'.
Perfusion Imaging
In cardiac MR perfusion imaging, the influx of intravenous gadolinium-based contrast
agent into the myocardium is analyzed under pharmacological stress and/or resting
conditions using single-shot gradient echo (GRE) sequences (flash, bSSFP, and GRE-EPI
hybrid sequences, the former due to the lowest artifact susceptibility most commonly
used [40]). Perfusion series are typically recorded in 3 short-axis and one long-axis plane
in order to cover as many of the 17 AHA segments as possible. In clinical routine,
the interpretation of regionally delayed arrival of contrast agent in perfusion series
is primarily performed qualitatively. Stress and resting perfusion series are analyzed
in order to differentiate between myocardial ischemia and a dark rim artifact mimicking
a subendocardial perfusion defect [41]
[42].
In addition to the potential for objective detection of regional perfusion deficits,
quantitative perfusion measurements offer the possibility to detect global or diffuse
myocardial perfusion changes, such as in non-obstructive coronary heart disease [7]
[8]
[43]. At present, the myocardial perfusion reserve index (MPRI) is the most established
cardiac MRI perfusion parameter. The MPRI is the ratio of the upslopes of the myocardial
signal intensity curves during first pass of contrast agent under stress and rest
([Fig. 8]). The rationale of this semi-quantitative parameter is that upslopes are a measure
of myocardial perfusion so that their quotient can be interpreted as a non-invasive
correlate to the coronary fractional flow reserve (FFR) [44]
[45]
[46]. A key limitation of the index, despite its conceptual simplicity, is the dependence
on the protocol parameters (e. g., amount of contrast agent, sequence parameters,
time between stress and resting perfusion acquisition), and the evaluation software
used [47].
Fig. 8 Assessment of the myocardial perfusion reserve index (MPRI). Motion corrected perfusion
images (left) can be employed to derive pixel-wise signal intensity time couses during
first pass of contrast agent (middle). Pixel-wise upslope maps during first pass of
contrast agent can be evaluated regionally or in AHA-segments, respectively (right).
MPRI is defined as stress-to-rest upslope ratio, which is reduced in the ischemic
AHA segment 10 compared to the non-ischemic AHA segment 7.
A more universal absolute quantification of myocardial perfusion requires the conversion
of signal intensity curves in the myocardium and left ventricular blood pool into
corresponding contrast agent concentration curves [8]. This results, however, in special requirements for data acquisition and/or contrast
agent application. Although the feasibility of even pixel-by-pixel perfusion quantification
has been demonstrated [44], this method is not yet clinically established [48].
Relaxation Time Mapping
Myocardial relaxation time mapping refers to the pixel-by-pixel estimation of magnetic
relaxation times (T1, T2, T2*) of the myocardium. Analogously to T1-, T2- or T2*-weighted
images, relaxation time maps display regional myocardial changes as regionally different
relaxation times. Their potential lies primarily in the fact that by segmentation
of the myocardium or myocardial evaluation regions, mean myocardial relaxation times
can be determined and compared with normal values [6]
[9]. Thus, not only regional but also diffuse and global myocardial changes can be identified
([Fig. 9]).
Fig. 9 Myocardial relaxation time maps. a Native T1-, T2, and T2* maps together with a post-contrast T1 map (T1-post). b Regional T1, T2, and ECV alterations in acute myocarditis. T1, T2, and ECV values
in involved lateral myocardium, as visualized also in the T2 weighted and late enhancement
images, are higher than in the non-involved septal myocardium. c Global T1 and ECV alterations in cardiac amyloidosis. Globally severely elevated
native T1 in absence of myocardial edema (normal myocardial T2) has been introduced
as superior diagnostic marker for cardiac amyloidosis, especially in case of atypical
late enhancement.
Cardiac relaxation time maps are calculated from image series with a varying sequence
parameter using known relationships between relaxation times and signal intensities.
Current mapping methods allow the image series to be captured within one breath hold
interval. For T1 mapping, the Modified Look-Locker Inversion recovery (MOLLI) sequence
is the most established method, where T1 values are obtained from single-shot bSSFP
image series with varying inversion time [49]
[50]
[51]. To estimate myocardial T2 times, for example, single-shot GRE image series with
varying T2 preparation times are used [51]
[52] and T2* times are finally calculated from segmented multiecho GRE images with different
echo times [6]
[52].
T1 and T2 Mapping
While in native myocardial T1 maps, myocardial fibrosis and myocardial edema are associated
with elevated T1 values, in T1 maps after contrast agent application (post-contrast
T1 maps), areas with increased extracellular space in myocardial fibrosis or necrosis
are associated with decreased T1 values corresponding to the higher contrast agent
distribution volume [49]
[53]. Post-contrast T1 values are determined not only by the morphological change of
the myocardium but also by the amount of contrast agent applied, the time after application
of the contrast agent and the contrast agent wash out behavior of the myocardium.
It is therefore common to calculate the extracellular volume fraction (ECV), which
is largely independent of the contrast agent kinetics [50]
[54] and reflects the relative distribution volume of the contrast agent in the myocardium
([Fig. 9]). Since myocardial edema significantly limits the association between elevated native
T1 and ECV values with myocardial fibrosis, it is recommended to interpret T1 and
ECV maps together with additional T2 maps in which myocardial edema present with elevated
T2 time [55]. While myocardial T1 and T2 time changes are not specific, they have been shown
to increase cardiac MRI-based diagnostic accuracy in myocarditis, amyloidosis, Anderson-Fabry
disease and cardiac iron storage disorders [6]
[56].
Although the acquisition and evaluation of myocardial relaxation time maps is simple,
the acquisition and evaluation technique must be taken into account in the interpretation
of the values ([Fig. 10]). Relaxation times determined pixel-by-pixel depend significantly on selected sequence
and protocol parameters; regional relaxation time averages can be falsified by partial
volume effects of the structures adjacent to the myocardium (blood, fat, pericardial
fluid) [57]
[58]. Accordingly, normal values identical to clinical routine mapping protocols must
be obtained, whereby acquisition and evaluation must be performed in a standardized
manner [6]
[50]. If the pixel-by-pixel overlay of the image series is limited by (movement) artifacts,
affected myocardial segments must be detected and excluded from the evaluation.
Fig. 10 Potential pitfalls in the interpretation and evaluation of relaxation time maps.
a Motion between images. Motion correction algorithms typically employed prior to map
calculations might fail to perfectly superpose pixel throughout image series (left).
Resulting motion artefacts might be difficult to be identified in maps but can be
recognized in motion corrected image series (yellow frame). b Partial volume of blood, fat, or pericardial fluid might impair average myocardial
relaxation times in regions-of-interests at myocardial borders [57]. For comparison of relaxation times with normal ranges, standardized evaluation
of maps is essential. c Pixels containing fat and myocardium might present with artificial T1 values in T1
maps [58]. As illustrated in the case of lipomatous metaplasia (arrows) T1 maps show low values
in pixels predominantly containing fat (lateral), but demonstrate artificially high
values in the septum and the lateral border zones. * indicates motion artefact as
described in a (compare T1-post map without motion artefact).
T2* Mapping
Reduced septal myocardial T2* times at 1.5 T are the reference standard method for
non-invasive diagnosis and grading of cardiac hemochromatosis [6]
[49]. For the severity classification of myocardial iron storage by T2* mapping, 1.5 T
T2* times of the basal septum over 20 ms are interpreted as normal, values less than
10 ms indicate severe myocardial iron storage [6]
[60]
[61]. The artifact susceptibility of the method and the lack of reference values for
3 T could contribute to establishing the reduction of myocardial T1 and T2 values
caused by cardiac iron storage diseases as new diagnostic markers [6].
Summary
Quantitative functional, phase contrast, and perfusion imaging, as well as relaxation
time mapping techniques give opportunity for assessment of a large number of quantitative
cardiac MRI parameters in clinical routine.
Application of these techniques allows for characterization of function, morphology
and perfusion of the heart beyond visual analysis of images, either in primary evaluation
and comparison to normal values or in patients’ follow-up and treatment monitoring.
However, with implementation of quantitative parameters in clinical routine, standardization
is of particular importance as different acquisition and evaluation strategies and
algorithms may substantially influence results, though not always immediately apparent.
