Introduction to Cardiovascular Magnetic Resonance: Technical Principles and Clinical Applications (2024)

1.1 Magnetic resonance

Magnetic resonance is a physical property of a nucleus that has an odd number of protons and/or neutrons. The most abundant nucleus of such kind in the body is the hydrogen proton (¹H). When a person is placed in a static magnetic field B₀, the magnetic moment μ of the hydrogen proton, also known as “spin”, rotates about the direction of B₀ (z direction) like a gyroscope with a fixed precession frequency. The precession frequency is called the Larmor frequency, and is proportional to the strength of B₀. If a group of spins rotate about B₀, these spins will form a net magnetization M₀ pointing in the +z direction. When a radiofrequency (RF) pulse of the same Larmor frequency is applied to these spins, the spins will absorb the energy from the RF pulse (resonance) and M₀ will be tilted away from the +z direction. The dynamics of the magnetization can be decomposed into two components changing in time independently; one component aligns with the +z direction and the other component lies in the plane perpendicular to the +z direction (xy plane). These two components of the magnetization are known as the longitudinal magnetization (M_z) and transverse magnetization (M_xy), respectively. When a RF coil transmits the RF pulse to the spins, M₀ starts to rotate away from the +z direction and eventually lies on the xy plane. Such a RF pulse is called a 90° RF pulse. If the RF pulse is applied by doubling the pulse duration or strength, the magnetization will tilted to the –z direction and becomes –M₀. Such a RF pulse is called 180° RF pulse or inversion pulse (Figure 1).

1.2 Relaxation

When the RF pulse is turned off, the magnetization gradually returns to its equilibrium state and becomes M₀ again. The process of returning to the equilibrium state is called relaxation. During relaxation, the magnitude of M_z increases toward the magnitude of M₀, whereas the magnitude of M_xy decreases toward zero. The relaxation process in M_z is called the longitudinal relaxation, and that in M_xy is called the transverse relaxation. The time it takes for M_z to increase from zero to 63% of the M₀ is known as T₁ (Figure 2), and the time for M_xy to decrease to 37% of M_xy at the beginning is known as T₂ (Figure 3). The length of T₁ is related to the energy exchange between proton spins and its surrounding molecules (spin-lattice interaction), whereas the length of T₂ is related to the energy exchange between any two proton spins (spin-spin interaction). The values of T₁ and T₂ depend on the characteristics of the tissue, and so they are unique inherent properties of the tissue. The differences in T1 and T2 in different tissues provide the source of image contrast in MRI.

1.3 Pulse sequence

A pulse sequence is a train of RF pulses that are used to produce MR signals. Different sequences can produce different signals from the tissue, depending on the imaging parameters and inherent T₁ and T₂ values. In other words, a pulse sequence can provide a distinct image contrast. Understanding the design of a pulse sequence will help us understand the mechanism of its image contrast. In this section, we will discuss two major types of pulse sequences that are commonly used in cardiovascular magnetic resonance (CMR).

1.3.1 Spin-echo (SE) sequence

In the SE sequence, a 90° RF pulse is first applied to rotate the magnetization M₀ from the +z direction to the xy plane. The transverse magnetization in the xy plane begins to decay due to the transverse relaxation. After a period of time, a 180° RF pulse is applied to refocus the decaying transverse magnetization and produce a signal, called ‘spin echo’ (Figure 4). There are two important imaging parameters in the SE sequence, namely echo time (TE) and repetition time (TR). TE is the time from the application of the 90° RF pulse to the formation of the spin echo, and TR is the time from the 90° RF pulse to the next 90° RF pulse. We can change the image contrast to be weighted toward T1 or T2 values or even toward the proton density by modifying the lengths of TR and TE. The rule of thumb is that short TR and short TE produce T1-weighted contrast, long TR and long TE produce T2-weighted contrast, and long TR and short TE produce proton-density-weighted contrast.

When the SE sequence is used in CMR, TR must coincide with the R-to-R interval under cardiac gating and TE must be very short to minimize the cardiac motion during data acquisition. This would make the scan time extremely long and render images with proton-density contrast only. To address this limitation, a fast SE sequence is developed. The fast SE sequence uses a 90° RF pulse followed by a rapid train of 180° RF pulses to produce a rapid train of spin echo signals. Instead of transmitting multiple 90° RF pulses with TR equaled to 1 R-to-R interval in the SE sequence, the fast SE sequence uses a single 90° RF pulse followed by a rapid train of 180° RF to acquire all the necessary echo signals within a single heartbeat (single shot) (Figure 5). This would inevitably blur the image due to cardiac motion and limited matrix resolution. If high spatial resolution is required, one could divide the single-shot mode into multiple shots and complete the data acquisition in a few heartbeats (Figure 6). To increase the image contrast, an effective way is to use inversion pulses before the 90° pulse (called pre-pulses) to suppress the signal from the blood or fat. Owing to high spatial resolution and adequate image contrast, the fast SE sequence is often used to acquire morphological information of the cardiovascular structures.

Open in a separate window

Figure 6

See Also

Mineralocorticoid receptor antagonist in heart failure: Past, present and future perspectivesCardiac MRI in diagnosis, prognosis, and follow-up of hypertrophic cardiomyopathy in children: current perspectivesCardiac MRI: Procedure DetailsRole of cardiovascular magnetic resonance in acute and chronic ischemic heart disease

The fast spin echo pulse sequence applies multiple 180° pulses following the 90° pulse to generate multiple spin echoes. Multiple lines of k space are filled within each heartbeat.

1.3.2 Gradient-echo (GRE) sequence

The GRE sequence uses a ‘weak’ RF pulse to tip the magnetization M₀ away from the +z direction by a small angle, and refocus the decaying transverse magnetization by a bipolar magnetic gradient. The echo signal produced by the GRE sequence is called the gradient echo or field echo (Figure 7). Compared to the standard SE sequence, the GRE sequence reduces the scan time by means of the two modifications mentioned above. First, it uses the small flip-angle RF pulses rather than 90° RF pulses to shorten the time required for longitudinal relaxation. Second, it uses bipolar refocusing gradients to avoid the time needed for the application of 180° RF pulses. These modifications allow TR in the GRE sequence to be as short as tens of milliseconds. By improving the switching rate of the bipolar gradients, a faster version of GRE sequence called fast (or turbo) GRE sequence has been developed in which TR can be shortened to a few milliseconds. The fast GRE sequence can acquire an image in a few hundred milliseconds, and so it allows acquisition of dynamic imaging for ventricular function, myocardial perfusion imaging or flow quantification.

1.4 Imaging

When the signal is produced by a pulse sequence, three spatially-encoding imaging gradients, namely slice-selection gradient (g_ss), phase-encoding (g_pe) and frequency-encoding gradient (g_fe) are consecutively played out to resolve the spatial location of the emitted signal. This task is called imaging. In theory, the MR signal, which has been encoded by the imaging gradients, is the Fourier transform of the object function in space. Therefore, the image of an object can be reconstructed by the Fourier transform of the signal in the spatially-encoding space (k space). In practice, when a RF pulse with a specific frequency band is applied, a slice-selection gradient g_ss is turned on at the same time. Since g_ss endows spins with different precession frequencies along the direction of g_ss, the RF pulse will only excite spins inside the slice of interest that have the same frequency, and make the M₀ within the slice rotate away from the +z direction. After the excitation of the RF pulse, the phase-encoding gradient g_pe and frequency-encoding gradient g_fe are successively turned on to encode the signal with the spatial information in y and in x coordinates, respectively (Figure 8). As the signals are collected under successive applications of g_pe and g_fe, they are registered at the appropriate positions in the spatially-encoding space (k space). Once the signals in the k space are completely registered, the image of an object is readily obtained by 2D Fourier transform of the signal in the k space.

1.5 Synchronization with electrocardiogram (ECG)

Unlike stationary organs, imaging of the beating heart should have a way to “freeze” the heart during data acquisition. Cardiac synchronization with ECG is a standard way to achieve this goal. There are two ECG synchronization methods in CMR. One is called the “prospective gating” method. This method uses a preceding ECG R-wave as a trigger to acquire a segment of k-lines in the k space within a short period of time; the length of the segment determines the temporal resolution of a cardiac phase. The same trigger-acquisition cycle is repeated in the following R-to-R intervals until all the necessary k-lines are collected. One should note that the acquisition window of the prospective gating method must be shorter than one R-to-R interval. If the acquisition window is longer than one R-to-R interval, the next trigger would not occur in the next R-wave and will skip to the third R-wave, and will double the acquisition time. To avoid prolonged acquisition, the acquisition window is often set to be around 80~90% of the average R-to-R interval. This, however, prohibits us from obtaining images at end diastole.

The other synchronization method, called “retrospective gating”, is to acquire data continuously and to retrospectively match the data to the corresponding positions of the ECG gating. Because data is reassigned over the k space according to the relative timing in the R-to-R interval, this synchronization method is capable of imaging the entire cardiac cycle.

2.1 Dark-blood MR imaging

The dark-blood technique uses double inversion recovery (IR) preparation pulses to null the blood signal, and delineates cardiac and vascular structures with the blood darkened. After the R-wave is detected, the first IR pulse (180° RF pulse) is applied to excite the spins of the whole imaging volume, followed immediately by the second slice-selective IR pulse to realign the spins in the specific imaged slice in the +z direction. The fast SE sequence is applied to collect SE signals when the longitudinal relaxation of the M_z in the blood passes zero (null) value. Therefore, the blood appears dark as compared to the myocardium in the acquired images. Triple IR technique is another dark blood technique which consists of the two IR pulses for dark blood preparation, followed by a third slice-selective IR pulse before the fast SE sequence to suppress the fat signal.

2.2 Clinical applications

Dark blood imaging is routinely used in the assessment of vascular abnormalities such as congenital anomalies, aortic or vascular disease, and myocardial abnormalities such as cardiomyopathy, myocarditis, and cardiovascular tumors. As shown in Figure 9, right ventricular lipoma is diagnosed based on the fact that the tumor shows high signal intensity on the dark-blood image and low signal intensity on the dark-blood dark-fat image.

3.1 Cine MRI

Current cine MRI uses a balanced steady-state free precession (bSSFP) technique combined with retrospective ECG gating. The bSSFP is a rapid GRE sequence in which the phase shift induced by the imaging gradients is exactly zero at the time of TR. This ensures high signal intensity of the images despite very short TR. In addition, the bSSFP sequence produces high tissue contrast between the myocardium and blood owing to its unique T1/T2 contrast and inflow effect. Because of these advantages, cine MRI can assess cardiac functions with high accuracy and reproducibility, and has been considered as a gold standard in the assessment of systolic function of the left ventricle.¹1 In addition to volumetric quantification, cine images acquired in two-chamber view, four-chamber view, short-axis view, and left ventricle/right ventricular inlet-outlet view also allow evaluation of the valvular insufficiency, outflow tract obstruction, dyssynchrony or asynchrony of the ventricular wall, or mobility of the cardiac tumors.

3.2 Ventricular function assessment

Global function of the left ventricle (LV) or right ventricle (RV) can be assessed with cine MRI in multiple short-axis views based on Simpson’s rule.² Ventricular volume can be estimated by summing multiple parallel sub-volumes in the short-axis view. This approach has been shown to be highly accurate and reproducible, but it requires software and human intervention to define the endocardial contour on each cine image. Once the endocardial contour of the LV or RV is defined, the area enclosed by each contour can be measured at each slice level. According to Simpson’s rule, ventricular volume at a particular cardiac phase can be estimated by summing the encircled areas of a stack of slices from apical to basal level and multiplying the slice thickness. Once the ventricular volumes are determined at all of the time frames in a cardiac cycle, we can obtain the volume-time curve of the ventricle (Figure 11). From the ventricular volume-time curve, the peak and valley values can be read off as the end-diastolic volume (EDV) and end-systolic volume (ESV), which are then used to compute the stroke volume (SV; SV = EDV – ESV) and ejection fraction (EF; EF = SV / EDV).³ The rate-of-volume-change curve is obtained by taking the time derivative of the volume-time curve (Figure 12). The minimal and maximal values can be read off from the rate-of-volume-change curve as the peak ejection rate (PER) and peak filling rate (PFR), respectively, indicating the systolic and diastolic functions. In addition, the time for deceleration (Tdec) is determined as the time interval between the maximal rate of early diastolic LV filling and the zero intercept of the deceleration slope. The Tdec can be used to indicate chamber stiffness.⁴ The LV mass is quantified by taking the difference between the ventricular volume measured from the epicardial contours and EDV, and multiplied by the density of the myocardium, 1.05 g/cc. The mass-to-volume ration is a useful index of LV geometric remodeling, and is calculated by the ratio of the LV mass divided by the EDV.⁵

Accurate quantification of the RV is important for congenital heart disease, such as Tetralogy of Fallot, transposition of great arteries and arrhythmogenic RV dysplasia.^6-8 The RV function estimated based on Simpson’s rule is usually recommended over the geometric assumption due to the complex geometry of the RV.

Cine images in short-axis view are often used for volumetric quantification. This approach is subject to an error due to ventricular shortening in the longitudinal direction. The longitudinal shortening mainly arises from the descent motion of the ventricular base during systole. Because of this motion, the number of ventricular slices at end systole should be fewer than that at end diastole. Without correcting for the longitudinal shortening, we may overestimate the ESV and underestimate the EF. To correct for the error, a simple way is to omit the slices that position in the atrium at end systole.

Besides global ventricular function, we can quantify regional wall thickening for each myocardial segment by calculating the ratio of the wall thickness at end diastole to that at end systole.

4.1 Late gadolinium enhancement MRI (LGE-MRI)

The most widely used MR contrast agent is the water soluble gadolinium (Gd) chelated agent. It is primarily in the intravascular space and permeates to the interstitial space of the tissue except brain (due to the blood brain barrier). The contrast agent undergoes exchange between the intravascular space and interstitial space and does not enter the intracellular space. After intravenous injection of Gd, both normal and abnormal myocardium will show different concentration curves of the Gd due to different kinetics. Compared to normal myocardium, abnormal myocardium such as infracted or scarred myocardium has larger interstitial space due to the loss of intact cardiomyocytes. With a time delay of approximately 10 minutes after Gd injection, abnormal myocardium will retain more contrast agent than normal myocardium. Since the Gd-chelated contrast agent is a contrast agent that predominantly shortens the longitudinal relaxation time (T1) of the proton spins, abnormal myocardium with increased Gd concentration will show signal enhancement on LGE-MRI images (hyperenhancement).

The LGE-MRI is performed using a T1-weighted rapid GRE sequence combined with an inversion-recovery pre-pulse to null the signal of normal myocardium (Figure 13). The images of this sequence show vivid contrast between normal myocardium (dark or no signal) and abnormal myocardium (bright or hyperenhancement). The sequence relies on operators to choose a correct inversion time (TI) among a range of TIs that best suppresses the signal from the normal myocardium (~250 ms). In practice, TI varies with time after Gd injection, and so operators may need to adjust TI several times during the acquisition of LGE-MRI. This will prolong the scan time. Recently, a new LGE sequence called phase-sensitive IR (PSIR) has been developed to address this problem. The PSIR sequence is designed in a way that it nulls the normal myocardium consistently over a range of TIs. Operators only need to use a default TI for the whole acquisition of LGE-MRI without adjusting it. Typically, one slice of 2D LGE-MRI is acquired in one breath-hold. With the advancement of parallel imaging, several slices can be acquired in one breath-hold. If patients cannot hold the breath, 3D LGE-MRI with respiratory navigator technique can be performed to cover the whole heart in approximately 5 minutes.

4.2 Clinical applications

LGE-MRI has become a standard clinical technique to visualize myocardial scarring. It allows us to identify spatial location and extent of the scar, the information of which may facilitate differential diagnosis of cardiomyopathy. Ischemic cardiomyopathy and non-ischemic cardiomyopathy have distinct etiologies, and have specific patterns of hyperenhancement.⁹ Therefore, we can distinguish ischemic cardiomyopathy from non-ischemic cardiomyopathy based on different enhancement patterns. In addition, specific enhancement patterns have been characterized in different etiologies of non-ischemic cardiomyopathy such as hypertrophic cardiomyopathy, arrhythmogenic RV dysplasia, Anderson-Fabry’s disease, amyloidosis, sarcoidosis and acute myocarditis (Figure 14).¹⁰ In patients with ischemic heart disease, it is important to distinguish dysfunction but viable tissue (hibernating myocardium) from the injured myocardium to regain contractility after revascularization.¹¹ In combination with stress-induced contractility reserve, LGE-MRI is one of the surrogates used to determine the hibernating myocardium.¹²

The severity of myocardial fibrosis or infarct is known to be related to ventricular arrhythmia and cardiac sudden death. Quantification of the hyperenhacement on LGE-MRI has been reported to be of value in predicting adverse outcomes in patients with non-ischemic cardiomyopathy.¹³ The most common quantification method is to define a remote zone which is normal myocardium away from the infarct tissue. A threshold of signal intensity is set at a value two standard deviations above the average signal (mean + 2SD) measured in the remote zone. Tissue with signal intensity above the threshold is considered as the infarct or scar tissue. The method has been applied to patients with systolic heart failure to identify conductive channels which are channels of viable myocardium inside the scar tissue connecting the surrounding viable myocardium. Presence of conductive channels in these patients was found to be predictive of ventricular tachyarrhythmia/fibrillation (VT/VF).¹⁴

4.3 Myocardial T1 mapping

As opposed to focal myocardial fibrosis or scarring which can be visualized on LGE-MRI, diffuse and interstitial myocardial fibrosis is not readily observable on LGE images because the signal intensity of the affected myocardium is not vividly enhanced. Instead of visualizing the hyperenhanced myocardium, quantification of T1 time may provide a means to assess the degree of myocardial fibrosis invisible to LGE-MRI. The modified Look-Locker inversion recovery (MOLLI) sequence is a novel pulse sequence that allows T1 quantification of the myocardium.¹⁵ Quantitative assessment of myocardial T1 time can be performed with or without administration of Gd, and has been considered to be an important biomarker in various myocardial diseases. Increased pre-contrast myocardial T1 time has been reported in several cardiomyopathies, such as myocarditis, amyloidosis, and hypertrophic cardiomyopathy.^16-18 In contrast, patients with iron overload myocardium and Anderson-Fabry’s disease show decreased pre-contrast myocardial T1 time.^19,20 Several studies have tried to use post-contrast myocardial T1 time to quantify diffuse myocardial fibrosis in patients with various cardiomyopathies.^21-23 However, the measurement of post-contrast myocardial T1 time is affected by several factors such as magnetic field strength, the timing of post-contrast MOLLI acquisition, the type of MOLLI scheme, the amount of contrast injected and the subject’s renal function.

Other studies proposed to measure the change in T1 time in pre-contrast vs. post-contrast conditions, the information from which allows us to calculate extracellular volume fraction (ECV) (Figure 15).²⁴ Myocardial ECV is defined as the ratio of the change in the myocardial T1 to the change in the blood T1. Lee et al.²⁵ compared myocardial ECV and T1 times with different types of MOLLI sequences at different post-contrast scanning times and different field strengths (1.5T and 3T). They found that myocardial ECV appeared to be stable and less affected by the testing variables described above. Recent studies have proven that increased ECV is associated with the severity of diffuse myocardial fibrosis in histology,²⁶ and with diastolic dysfunction in patients with heart failure and preserved ejection fraction.²⁷ Therefore, myocardial ECV is more favorable than post-contrast myocardial T1 time to estimate the extracellular matrix expansion in diffuse myocardial fibrosis.

Open in a separate window

Figure 15

The flowchart of quantifying the extracellular volume fraction (ECV). For T1 maps, the regions of interest (ROI) in the blood and the myocardium of the LV are segmented manually in the central area of the LV cavity and the septal myocardium, respectively. If the septal myocardium shows regional hyperenhancement on the late gadolinum enhancement (LGE) images, the region of interest of the myocardium is redrawn in other unenhanced myocardial region. The ECV is calculated using the ratio of the change in relaxation rate (R₁ = 1/T₁) in the myocardium to that in the blood and multipled by (1-hematocrit).

4.4 Myocardial T2 and T2* mapping

Myocardial T2 mapping is a technique used to reconstruct a parametric image based on the T2 value measured in each voxel. The accumulation of water in the myocardium is associated with different types of pathology, such as acute myocardial infarction, myocarditis and graft rejection. Given a long transverse relaxation time of water protons, myocardial T2 mapping appears to be promising to detect intramyocardial water and even quantify myocardial edema. Verhaert et al. reported that T2 mapping may be useful to identify myocardial regions at risk and microvascular obstruction.²⁸

T2* relaxation refers to the rapid decay of transverse magnetization caused by local magnetic field inhom*ogeneity. Magnetic filed inhom*ogeneity comes from the inhom*ogeneity of the static magnetic field or from the differences in magnetic susceptibility between neighboring tissues, such as air-tissue interfaces, metallic implants, paramagnetic contrast agents or iron deposition. Specifically, myocardial T2* mapping is sensitive to tissue iron content, and so it is widely used to quantify the degree of iron deposition in the myocardium in patients with thalassemia major and repeated blood transfusion.²⁹

5.1 First-pass contrast-enhanced MRI (FPCE-MRI)

Myocardial perfusion imaging is an imaging technique designed to assess the microcirculation of the myocardium. The technique uses a T1-weighted fast GRE sequence to perform dynamic imaging right after bolus injection of Gd. A series of dynamic images are acquired during the first-pass when Gd passes through the right heart, lungs, left heart, aorta and flows into the myocardium via the coronary arteries. When Gd flows into the myocardium, the T1 time shortens and the myocardium is enhanced according to the concentration of Gd. Typically, the temporal resolution of FPCE-MRI is kept at one cardiac cycle. Given the speed of current fast GRE sequence, only 3 to 5 short-axis slices at the basal, mid, and apical levels are acquired within each heartbeat. The acquisition lasts about one minute, yielding 60 to 80 time frames per slice. The acquisition in this time window can cover the complete history of Gd in the first pass and early equilibrium state.

Myocardial perfusion is usually performed under rest and stress states. In contrast to the rest state, the stress state is induced pharmacologically by a vasodilator such as adenosine or dipyridamole, which causes hyperemia of the myocardium and increases perfusion. The ratio of perfusion at stress to that at rest is defined as myocardial perfusion reserve (MPR). In normal myocardium, MPR is above 2, and ischemia is considered if MPR is below 1.5. To ensure complete washout of Gd from the myocardium, the stress and rest tests should be separated by approximately 10 minutes.

5.2 Qualitative assessment

Qualitative assessment is to visualize (eye-balling) perfusion defects based on relatively reduced enhancement in the ischemic myocardium. A full dose of Gd (0.1 mmole/kg of body weight) is typically required to increase the detectability. However, Gd-related susceptibility artifacts becomes obvious in the ventricular cavity with concentrated Gd during the first pass. The artifact is located at the blood-myocardium interface, causing signal drop in the subendocardium. This artifact is sometimes misdiagnosed as myocardial ischemia. Therefore, qualitative measurement highly depends on the reader’s experience.

5.3 Quantitative assessment

Myocardial perfusion can be assessed quantitatively by analyzing the dynamic change of the signal intensity over time. Analysis of myocardial perfusion image starts with segmenting the LV myocardium and cavity. The LV myocardium at three short-axis levels is divided into 16 equiangular segments (not including the apical cap) according to the guidelines provided by the American Heart Association.³⁰ The mean signal intensity in each myocardial segment and LV cavity is measured at each time point to obtain the signal-intensity-time curve (SI curve). Myocardial perfusion can be assessed from the SI curves semi-quantitatively or quantitatively.

5.3.1 Semi-quantitative analysis

Myocardial perfusion can be assessed by characterizing the SI curves in terms of maximal up-slope, peak value, time to peak and area under the curve. Among the above parameters, the maximal up-slope and its ratio (MPR) have been widely adopted for semi-quantitative analysis of myocardial perfusion. The time window of the first pass is determined from the SI curve of the LV cavity; the time of the onset of contrast arrival and the time of the onset of recirculation are defined as the beginning and end of the time window, respectively. The SI curve of the first pass in the myocardial segments at the same level as the LV cavity is fitted by a gamma-variate function (Figure 16). The maximal up-slope is calculated from the peak value of the time derivative of the fitting function in the myocardium and LV cavity. Although semi-quantitative methods provide little insight to the underlying physiology, they are relatively easy to use and can assess MPR with high accuracy and reproducibility. MPR derived from the maximal up-slope has been validated by empirical observations in radiolabeled microspheres measurement in a closed-chest dog model.³¹ Several studies have demonstrated that semi-quantitative methods reveal strong diagnostic power to detect patients with coronary artery disease. Al-Saadi et al. and Nagel et al. used upslope ratio to detect coronary artery stenosis, reporting that the sensitivity, specificity and diagnostic accuracy were 90%, 83%, 87% and 88%, 90%, 89%, respectively.³²

5.3.2 Quantitative analysis

Quantitative analyses of myocardial perfusion can be broadly divided into two categories: model-dependent and model-independent analyses.³³ For model-dependent analysis, one should specify functional compartments (vessel, interstitium and myocyte) in the myocardial tissue and how the tracers (contrast agent) exchange between these compartments. As mentioned previously, Gd-chelated agent is a water-soluble molecule and does not enter the intracellular space. One can reduce the functional compartments to vascular and interstitial compartments and simplify the model into a two-compartment model.³⁴ Compartment analysis has been widely used to quantify regional myocardial blood flow for myocardial perfusion imaging. One should note that the compartment analysis assumes that the tracers mix very quickly within each compartment. The assumption is made to neglect concentration gradients within the compartment and heterogeneous flows in different vessels.

In 1992, Diesbourg and his colleagues proposed a kinetic model to quantify myocardial blood flow and extracellular volume for ischemic and infarct myocardium.³⁵ They suggested that myocardial blood flow (F) and extracellular space (V) can be calculated by fitting the contrast concentration as a function of time in the myocardium [Gd-DTPA]_T(t) by the following equation:

where [Gd-DTPA]_A(t) is the arterial blood concentration of Gd-DTPA as a function of time, E is the extraction efficiency and λ is the partition coefficient, which is equivalent to V.

Currently, most methods applied to quantify myocardial blood flow with MRI or other modalities such as positron emission tomography have relied on tracer kinetic models of varying complexity and physiological realism. However, the use of models raises questions as to which model is most appropriate, and what is the best optimization strategy to determine some of the model parameters from the measured data.

In 2002, Jerosch-Herold first introduced the model-independent analysis combined with Tikhonov’s regularization to determine myocardial blood flow based on the central volume principle.³⁶ According to the central volume principle, the tissue residue function [C_T(t)] is equal to the arterial input function [C_A(t)] and convolves with the impulse response function [R(t)].

C_T (t) = C_A (t) ⊗ R(t)

The initial height of the impulse response function is equal to the blood flow (Figure 17). This property is independent of the vascular or compartmental structures inside the region of interest (ROI) of the myocardium or the permeability of the vessel wall. Since Gd-chelated agent is an extravascular contrast agent, it leaks from the capillaries and enters the interstitial space by diffusion and convection. The contrast agent redistributes within the ROI, and it contributes little to the flow-in and flow-out of the tissue ROI because the transport process of diffusion and convection is much slower than the blood flow. Therefore, measurement of myocardial blood flow can be achieved directly from the impulse response function.

Open in a separate window

Figure 17

The flow chart of perfusion quantification analysis. The analysis starts with alignment of the perfusion images followed by surface coil signal correction. After segmentation of the left ventricle (LV), the time curves of mean signal intensity (SI; AU, arbitrary unit) in myocardium (Myo, myocardium) and in the LV cavity (AIF, arterial input function) are computed, and normalized by the area under the curve of the AIF. The residue impulse response (ml/sec/g) is calculated by the general singular value deconvolution (GSVD) of Myo and AIF, followed by Tikhonov regularization and determination from the L-curve criterion. For validation, the resulting impulse response is convolved with the AIF to compare with the time curve of the myocardium (Myo). After obtaining the impulse response, the myocardial perfusion (ml/min/g) is determined by the initial height of the impulse response.

5.4 Clinical applications

FPCE MRI has been shown to have better diagnostic accuracy in coronary artery disease than the other imaging modalities.³⁷ With exquisite spatial resolution of FPCE MRI, Lee et al. demonstrated that the transmural perfusion gradient can be identified from flow-limiting stenosis during global coronary vasodilatation.³⁸ There are plenty of researches using FPCE MRI to discriminate hibernating myocardium by myocardial perfusion reserve, to quantify severity of atherosclerosis, to evaluate vasodilator response of patients with hypertrophic cardiomyopathy, to study perfusion changes in patients with chronic myocardial infarction after cardiac rehabilitation or to assess perfusion of residual viable myocardium after percutaneous intervention.³⁹

6.1 Phase-contrast cine

Phase-contrast cine acquires a series of images within a cardiac cycle with cardiac gating similar to cine MRI. In addition to magnitude images, phase images are acquired to indicate flow velocity. This is achieved by using a bipolar velocity-encoding gradient added to the cardiac-gated GRE sequence to measure the phase shift caused by the flowing blood. The phase shift is proportional to the flow velocity along the direction of the bipolar gradient. In phase-contrast cine, the stationary tissue appears intermediate signal intensity (gray) on the phase contrast image, while the flowing blood would appear white or black signal intensity, depending on the direction of the flow relative to the velocity-encoding gradients. The magnitude of the velocity-encoding gradient should be adjusted according to the estimated maximal velocity of the target vessel. Otherwise, phase-contrast cine will suffer from aliasing artifact if the gradient is too high or insufficient velocity contrast if the gradient is too low. If the gradient is applied perpendicular to the imaging plane, the velocity of the “through plane” flow is measured (Figure 18). This is the most commonly acquired setting in a clinical examination. Phase-contrast technique can also be applied to measure flow velocity in three dimensions. The acquired data can be computed to generate virtual stream lines which represent blood flow within a vessel or cardiac chamber.^40,41

6.2 Clinical applications

Phase-contrast cine is primarily used for flow quantification of great vessels. The blood flow of a vessel can be determined by the averaged velocity within the vessel lumen multiplied by the lumen area. The total blood flow in a cardiac cycle can be determined by summing up the blood flows at all the cardiac phases. Regurgitant flow can be determined from the flow-time curve to quantify the severity of valvular regurgitation. Phase-contrast cine has been applied to measure pulse wave velocity of the descending aorta and was found to be associated with the elasticity of the aortic wall.⁴² It has also been used to estimate the severity of pulmonary hypertension.⁴³ In clinical practice, phase-contrast cine is used to quantify shunting flow in congenital heart disease, regurgitation fraction in valvular heart disease or flow within a conduit.

7.1 Contrast-enhanced MR angiography

Contrast-enhanced MR angiography (CE-MRA) uses Gadolinium chelate contrast agent to shorten T1 value of blood. With the employment of spoiled gradient echo sequence using short TR, the vessels will appear bright because of Gadolinium, while the background tissue signal is suppressed due to a saturation effect. The vessels would appear more distinct if further mask subtraction with pre-contrast MRA images are employed.⁴⁴ Power injector can also be used to increase contrast concentration and vessel conspicuity.⁴⁵

In most circ*mstances, we want to image only the target arteries without contamination from the veins. Therefore, the image timing and scan duration are thus important. The timing of contrast arrival is affected by patient age, body habitus, cardiac output, etc. The simplest way to set image timing is best guess along with multiphasic acquisitions. More precise methods include preliminary timing scan,⁴⁵ and real-time monitoring (MR fluoroscopy) during bolus injection with elliptical centric view ordering of k space.^46,47 The scan duration of CE-MRA generally lasts no more than a breath-hold to avoid respiratory artifact as well as venous enhancement. With the development of high-field (such as 3T) MR, parallel imaging technique, and high performance gradient coils, it is now possible to acquire high resolution, large FOV, three-dimensional CE-MRA images with reasonable scan time and good SNR, even with lower contrast dosage.⁴⁸ With this image data set, multiplanar reconstruction, maximal intensity projection, and volume rendered images can be post-processed for viewing and analysis. CE-MRA is useful to for imaging suspected aortic dissection, aneurysm, arteritis, anomaly, pulmonary thromboembolism,⁴⁹ and aortic graft patency (Figure 19).

Time-resolved CE-MRA further decreases the scan time of each image with data sharing technique and compromise in spatial resolution, usually in the slice direction. It can display multiple time frames of volumetric images with temporal resolution of 1-2 seconds to demonstrate vascular flow dynamics. It therefore has no constraints of scan timing and duration as previously described in CE-MRA. Such technique is useful in diseases such as aortic dissection, vascular malformation, cardiac shunting, and severe stenosis such as subclavian steal syndrome.

7.2 Coronary MR angiography

The ultimate goal of coronary MR angiography is to diagnose luminal stenosis. The goal is challenging because the diameter of a coronary artery is no more than 5 mm, and the arteries always move with cardiac and respiratory motion. In order to delineate the lumen of the coronary arteries, the spatial resolution needs to be 1 mm at least, and the motion-related artifacts should be minimized as much as possible. Current imaging technique uses a 3D GRE sequence with fat-saturation and synchronous ECG- and respiratory-gating (Figure 20). The respiratory gating uses one-dimensional navigator echo for continuous diaphragm tracing. The scanning of the whole heart takes 5 to 10 minutes under free breathing. Coronary MR angiography is suitable for identifying coronary artery openings and morphological abnormality such as coronary artery aneurysm or fistula. It has been shown to be able to diagnose three-vessel disease and main artery disease.⁵⁰ However, the spatial resolution is still inadequate and image quality is easily affected by variations in respiratory movement and cardiac cycle. Efforts are required to improve the technologies of coronary MR angiography before it can be used in routine clinical examination.

1. Leong DP, De Pasquale CG, Selvanayagam JB. Heart failure with normal ejection fraction: the complementary roles of echocardiography and CMR imaging. JACC Cardiovasc Imaging. 2010; 3(4):409–420. [PubMed] [Google Scholar]

2. Brenner DE, Whitley NO, Houk TL, et al. Volume determinations in computed tomography. JAMA. 1982;247(9):1299–1302. [PubMed] [Google Scholar]

3. Tseng WY, Liao TY, Wang JL. Normal systolic and diastolic functions of the left ventricle and left atrium by cine magnetic resonance imaging. J Cardiovasc Magn Reson. 2002;4(4):443–457. [PubMed] [Google Scholar]

4. Ohno M, Cheng CP, Little WC. Mechanism of altered patterns of left ventricular filling during the development of congestive heart failure. Circulation. 1994;89(5):2241–2250. [PubMed] [Google Scholar]

5. Gaasch WH, Delorey DE, St John Sutton MG, et al. Patterns of structural and functional remodeling of the left ventricle in chronic heart failure. Am J Cardiol. 2008;102(4):459–462. [PubMed] [Google Scholar]

6. Nollert G, Fischlein T, Bouterwek S, et al. Long-term survival in patients with repair of tetralogy of Fallot:36-year follow-up of 490 survivors of the first year after surgical repair. J Am Coll Cardiol. 1997;30(5):1374–1383. [PubMed] [Google Scholar]

7. Jimenez-Juan L, Joshi SB, Wintersperger BJ, et al. Assessment of right ventricular volumes and function using cardiovascular magnetic resonance cine imaging after atrial redirection surgery for complete transposition of the great arteries. Int J Cardiovasc Imaging. 2013;29(2):335–342. [PubMed] [Google Scholar]

8. Saguner AM, Vecchiati A, Baldinger SH, et al. Different prognostic value of functional right ventricular parameters in arrhythmogenic right ventricular cardiomyopathy/dysplasia. Circ Cardiovasc Imaging. 2014;7(2):230–239. [PubMed] [Google Scholar]

9. Mahrholdt H, Wagner A, Judd RM, et al. Delayed enhancement cardiovascular magnetic resonance assessment of non-ischaemic cardiomyopathies. Eur Heart J. 2005;26(15):1461–1474. [PubMed] [Google Scholar]

10. Jackson E, Bellenger N, Seddon M, et al. Ischaemic and non-ischaemic cardiomyopathies--cardiac MRI appearances with delayed enhancement. Clin Radiol. 2007;62(5):395–403. [PubMed] [Google Scholar]

11. Rizzello V, Poldermans D, Schinkel AF, et al. Long term prognostic value of myocardial viability and ischaemia during dobutamine stress echocardiography in patients with ischaemic cardiomyopathy undergoing coronary revascularisation. Heart. 2006;92(2):239–244. [PMC free article] [PubMed] [Google Scholar]

12. Wellnhofer E, Olariu A, Klein C, et al. Magnetic resonance low-dose dobutamine test is superior to SCAR quantification for the prediction of functional recovery. Circulation. 2004;109(18):2172–2174. [PubMed] [Google Scholar]

13. Wu KC, Weiss RG, Thiemann DR, et al. Late gadolinium enhancement by cardiovascular magnetic resonance heralds an adverse prognosis in nonischemic cardiomyopathy. J Am Coll Cardiol. 2008;51(25):2414–2421. [PMC free article] [PubMed] [Google Scholar]

14. Lin LY, Su MY, Chen JJ, et al. Conductive channels identified with contrast-enhanced MR imaging predict ventricular tachycardia in systolic heart failure. JACC Cardiovasc Imaging. 2013;6(11):1152–1159. [PubMed] [Google Scholar]

15. Messroghli DR, Radjenovic A, Kozerke S, et al. Modified Look-Locker inversion recovery (MOLLI) for high-resolution T1 mapping of the heart. Magn Reson Med. 2004;52(1):141–146. [PubMed] [Google Scholar]

16. Puntmann VO, Voigt T, Chen Z, et al. Native t1 mapping in differentiation of normal myocardium from diffuse disease in hypertrophic and dilated cardiomyopathy. JACC Cardiovasc Imaging. 2013;6(4):475–484. [PubMed] [Google Scholar]

17. Ferreira VM, Piechnik SK, Dall’Armellina E, et al. T(1) mapping for the diagnosis of acute myocarditis using CMR:comparison to T2-weighted and late gadolinium enhanced imaging. JACC Cardiovasc Imaging. 2013;6(10):1048–1058. [PubMed] [Google Scholar]

18. Karamitsos TD, Piechnik SK, Banypersad SM, et al. Noncontrast T1 mapping for the diagnosis of cardiac amyloidosis. JACC Cardiovasc Imaging. 2013;6(4):488–497. [PubMed] [Google Scholar]

19. Sado DM, White SK, Piechnik SK, et al. Identification and assessment of Anderson-Fabry disease by cardiovascular magnetic resonance noncontrast myocardial T1 mapping. Circ Cardiovasc Imaging. 2013;6(3):392–398. [PubMed] [Google Scholar]

20. Pedersen SF, Thrysoe SA, Robich MP, et al. Assessment of intramyocardial hemorrhage by T1-weighted cardiovascular magnetic resonance in reperfused acute myocardial infarction. J Cardiovasc Magn Reson. 2012;14:59. [PMC free article] [PubMed] [Google Scholar]

21. Mewton N, Liu CY, Croisille P, et al. Assessment of myocardial fibrosis with cardiovascular magnetic resonance. J Am Coll Cardiol. 2011;57(8):891–903. [PMC free article] [PubMed] [Google Scholar]

22. Iles L, Pfluger H, Phrommintikul A, et al. Evaluation of diffuse myocardial fibrosis in heart failure with cardiac magnetic resonance contrast-enhanced T1 mapping. J Am Coll Cardiol. 2008; 52(19):1574–1580. [PubMed] [Google Scholar]

23. Mascherbauer J, Marzluf BA, Tufaro C, et al. Cardiac Magnetic resonance post-contrast T1 time is associated with outcome in patients with heart failure and preserved ejection fraction. Circ Cardiovasc Imaging. 2013 [PubMed] [Google Scholar]

24. Kellman P, Wilson JR, Xue H, et al. Extracellular volume fraction mapping in the myocardium, part 1: evaluation of an automated method. J Cardiovasc Magn Reson. 2012;14:63. [PMC free article] [PubMed] [Google Scholar]

25. Lee JJ, Liu S, Nacif MS, et al. Myocardial T1 and extracellular volume fraction mapping at 3 tesla. J Cardiovasc Magn Reson. 2011;13:75. [PMC free article] [PubMed] [Google Scholar]

26. Flett AS, Hayward MP, Ashworth MT, et al. Equilibrium contrast cardiovascular magnetic resonance for the measurement of diffuse myocardial fibrosis: preliminary validation in humans. Circulation. 2010;122(2):138–144. [PubMed] [Google Scholar]

27. Su MY, Lin LY, Tseng YH, et al. CMR-verified diffuse myocardial fibrosis is associated with diastolic dysfunction in HFpEF. JACC Cardiovasc Imaging. 2014;7(10):991–997. [PubMed] [Google Scholar]

28. Verhaert D, Thavendiranathan P, Giri S, et al. Direct T2 quantification of myocardial edema in acute ischemic injury. JACC Cardiovasc Imaging. 2011;4(3):269–278. [PMC free article] [PubMed] [Google Scholar]

29. Pennell DJ. T2* magnetic resonance and myocardial iron in thalassemia. Ann N Y Acad Sci. 2005;1054:373–378. [PubMed] [Google Scholar]

30. Cerqueira MD, Weissman NJ, Dilsizian V, et al. Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart. A statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. Circulation. 2002; 105(4):539–542. [PubMed] [Google Scholar]

31. Wilke N, Simm C, Zhang J, et al. Contrast-enhanced first pass myocardial perfusion imaging: correlation between myocardial blood flow in dogs at rest and during hyperemia. Magn Reson Med. 1993;29(4):485–497. [PubMed] [Google Scholar]

32. Al-Saadi N, Nagel E, Gross M, et al. Noninvasive detection of myocardial ischemia from perfusion reserve based on cardiovascular magnetic resonance. Circulation. 2000;101(12):1379–1383. [PubMed] [Google Scholar]

33. Jerosch-Herold M. Quantification of myocardial perfusion by cardiovascular magnetic resonance. J Cardiovasc Magn Reson. 2010;12:57. [PMC free article] [PubMed] [Google Scholar]

34. Kroll K, Wilke N, Jerosch-Herold M, et al. Modeling regional myocardial flows from residue functions of an intravascular indicator. Am J Physiol. 1996;271(4 Pt 2):H1643–H1655. [PMC free article] [PubMed] [Google Scholar]

35. Diesbourg LD, Prato FS, Wisenberg G, et al. Quantification of myocardial blood flow and extracellular volumes using a bolus injection of Gd-DTPA: kinetic modeling in canine ischemic disease. Magn Reson Med. 1992;23(2):239–253. [PubMed] [Google Scholar]

36. Jerosch-Herold M, Swingen C, Seethamraju RT. Myocardial blood flow quantification with MRI by model-independent deconvolution. Med Phys. 2002;29(5):886–897. [PubMed] [Google Scholar]

37. Nagel E, Klein C, Paetsch I, et al. Magnetic resonance perfusion measurements for the noninvasive detection of coronary artery disease. Circulation. 2003;108(4):432–437. [PubMed] [Google Scholar]

38. Lee DC, Simonetti OP, Harris KR, et al. Magnetic resonance versus radionuclide pharmacological stress perfusion imaging for flow-limiting stenoses of varying severity. Circulation. 2004;110(1):58–65. [PubMed] [Google Scholar]

39. Su MY, Lee BC, Wu YW, et al. Perfusion of residual viable myocardium in nontransmural infarct zone after intervention: MR quantitative myocardial blood flow measurement. Radiology. 2008;249(3):820–828. [PubMed] [Google Scholar]

40. Kilner PJ, Yang GZ, Wilkes AJ, et al. Asymmetric redirection of flow through the heart. Nature. 2000;404(6779):759–761. [PubMed] [Google Scholar]

41. Frydrychowicz A, Arnold R, Hirtler D, et al. Multidirectional flow analysis by cardiovascular magnetic resonance in aneurysm development following repair of aortic coarctation. J Cardiovasc Magn Reson. 2008;10:30. [PMC free article] [PubMed] [Google Scholar]

42. Yu HY, Peng HH, Wang JL, et al. Quantification of the pulse wave velocity of the descending aorta using axial velocity profiles from phase-contrast magnetic resonance imaging. Magn Reson Med. 2006;56(4):876–883. [PubMed] [Google Scholar]

43. Kang KW, Chang HJ, Yoo YP, et al. Cardiac magnetic resonance-derived right ventricular outflow tract systolic flow acceleration: a novel index of right ventricular function and prognosis in patients with pulmonary arterial hypertension. Int J Cardiovasc Imaging. 2013;29(8):1759–1767. [PubMed] [Google Scholar]

44. Ruehm SG, Nanz D, Baumann A, et al. 3D contrast-enhanced MR angiography of the run-off vessels: value of image subtraction. J Magn Reson Imaging. 2001;13(3):402–411. [PubMed] [Google Scholar]

45. Earls JP, Rofsky NM, DeCorato DR, et al. Breath-hold single-dose gadolinium-enhanced three-dimensional MR aortography:usefulness of a timing examination and MR power injector. Radiology. 1996;201(3):705–710. [PubMed] [Google Scholar]

46. Riederer SJ, Bernstein MA, Breen JF, et al. Three-dimensional contrast-enhanced MR angiography with real-time fluoroscopic triggering: design specifications and technical reliability in 330 patient studies. Radiology. 2000;215(2):584–593. [PubMed] [Google Scholar]

47. Wilman AH, Riederer SJ. Performance of an elliptical centric view order for signal enhancement and motion artifact suppression in breath-hold three-dimensional gradient echo imaging. Magn Reson Med. 1997;38(5):793–802. [PubMed] [Google Scholar]

48. Tomasian A, Salamon N, Lohan DG, et al. Supraaortic arteries:contrast material dose reduction at 3.0-T high-spatial-resolution MR angiography--feasibility study. Radiology. 2008;249(3):980–990. [PubMed] [Google Scholar]

49. Kalb B, Sharma P, Tigges S, et al. MR imaging of pulmonary embolism: diagnostic accuracy of contrast-enhanced 3D MR pulmonary angiography,contrast-enhanced low-flip angle 3D GRE, and nonenhanced free-induction FISP sequences. Radiology. 2012; 263(1):271–278. [PubMed] [Google Scholar]

50. Kim WY, Danias PG, Stuber M, et al. Coronary magnetic resonance angiography for the detection of coronary stenoses. N Engl J Med. 2001;345(26):1863–1869. [PubMed] [Google Scholar]