Advances in Parametric Mapping with Cardiac Magnetic Resonance Imaging

Introduction

Cardiac magnetic resonance imaging (CMR) has emerged as a mature imaging modality and is considered the gold standard for quantification of myocardial mass, volumes and function.(1) Myocardial tagging and phase-contrast velocity imaging have been in use for over 20 years providing quantification of myocardial function and flow. While CMR has been used for tissue characterization for many years, it is only with recent improvements in MRI scanner technology and parallel imaging reconstruction techniques that parametric mapping of perfusion or magnetic relaxation properties (T1, T2, and T2*) has become clinically feasible within a single breath-hold. To generate a parametric map of perfusion or relaxation times, multiple images of the same region of the myocardium are acquired with different sensitivity to the parameter of interest, and the signal intensities of these images are fit to a model which describes the underlying physiology or relaxation parameters. The parametric map is an image of the fitted perfusion parameters or relaxation times. Parametric mapping techniques are extending the range of unique tissue parameters that can be measured by CMR. Quantitative perfusion imaging techniques measure myocardial perfusion and perfusion reserve (MPR), demonstrating improved abilities to assess the extent of myocardial ischemia in multi-vessel disease.(2) Pre and post-contrast T₁ mapping techniques can quantify diffuse myocardial fibrosis in hypertrophic and infiltrative cardiomyopathies.(3–5) T₂ mapping techniques are improving quantification of myocardial edema.(6,7) Imaging of the T₂* relaxation time which is sensitive to magnetic field inhomogeneity has become clinically useful in evaluation of myocardial iron overload and assessment of chelation therapy.(8,9) This article will review the current imaging techniques, emerging applications, as well as the future potential and limitations of CMR for parametric mapping of the myocardium. It will specifically focus of assessment of myocardial perfusion, and magnetic relaxation times (T1, T2, T2*).

Quantification of Myocardial Perfusion

Most clinical perfusion imaging techniques only assess relative differences in myocardial perfusion. The SPECT literature has established that measuring relative perfusion results in reduced sensitivity for the detection of left-main and three vessel coronary artery disease (CAD) as there may be no reference area with normal perfusion.(10,11) Furthermore, there is growing interest in the ability to assess microvascular dysfunction in women and patients with diabetes or chest pain and non-obstructive CAD.(12) PET is considered the current gold standard technique for quantitative perfusion imaging. Studies have shown that quantification of myocardial PET studies improves diagnostic accuracy and provides a useful adjunct to assessment of regional perfusion abnormalities. Abnormal MPR by PET in the absence of obstructive CAD is associated with adverse cardiovascular outcomes.(13) PET has a few disadvantages including high cost, limited ability to obtain tracers, poor spatial resolution, and exposure of patients to ionizing radiation.(14) CMR has demonstrated potential for clinical quantitative perfusion imaging and would overcome some of the aforementioned limitations of PET for this application.(15)

CMR Techniques for Quantitative Perfusion Analysis

In order to perform absolute quantification of myocardial blood flow, a quantifiable relationship must exist between the signal intensity changes in the image and underlying blood flow. First pass myocardial perfusion techniques utilize gadolinium (Gd) chelates that shorten the T₁ relaxation time of water (recovery of longitudinal magnetization) in proportion to the concentration of Gd using saturation-recovery (SR) pulse sequences to impart strong T₁-weighted (W) image contrast. In regions with normal blood flow there is an increased concentration of Gd resulting in a bright signal in the perfusion image, whereas in regions of reduced perfusion there are lower concentrations of gadolinium resulting in a lower signal intensity. Typically 3–5 slice positions are imaged in a sequential fashion during each heart-beat over the 30–60 seconds of the first-pass of the contrast agent. Multiple pulse sequences have been used for perfusion imaging and each have their advantages and disadvantages.(14) Parallel imaging with acceleration factors of 2–3 are routinely used (16) and multiple highly accelerated techniques have been evaluated in clinical studies. (17) More recent techniques which exploit the spatial-temporal correlations in the series of first-pass perfusion imaging using techniques such as kt-SENSE are enabling true 3D coverage of the ventricle with high spatial resolution.(18) Compressed-sensing techniques, which rely on the compressibility of image data, or image sparsity, hold potential for even more acceleration of perfusion image acquisition.(19) While any of these techniques can be used for quantification of myocardial perfusion, care must be taken to make sure that the techniques used do not disrupt the temporal evolution of the signal intensities.

In order to quantify myocardial perfusion, two essential components must be measured accurately: (1) the arterial input function (AIF) which quantifies the concentration of Gd delivered from the left ventricle and aorta as a function of time and (2) the tissue function (TF) which is related to the accumulation of contrast agent within the myocardium. Quantification of the AIF requires special attention as the high concentration of Gd in the blood pool results in significant signal saturation resulting in improper determination of the AIF. This problem can be overcome using several different strategies. One approach is using a lower dose of contrast agent; however, this results in images with lower signal-to-noise ratio (SNR). Another is the “dual bolus” or “pre-bolus” technique where a low dose of Gd is used for quantification of the AIF and a second higher dose of contrast is used to determine the TF.(20) This technique accurately quantifies both the AIF and TF, but it requires 2 boluses with different contrast concentrations which may be impractical. A third approach involves utilization of a “dual sequence” approach which uses a single high dose of contrast.(21) A low-resolution, low T₁ sensitivity acquisition is performed in each heart beat to quantify the AIF, while the standard pulse sequence is used to quantify the TF.

Once an accurate AIF and TF have been obtained, there are two main techniques to quantify perfusion, multi-compartment kinetic modeling and Fermi-function deconvolution. With compartmental modeling, the forward flux of Gd from the blood to the myocardium (K_trans) is taken to represent absolute myocardial blood flow.(22) An important factor which is usually neglected, is that the extraction of Gd, which is flow-dependent is required to convert K_trans into a measurement of absolute perfusion. Also, the fact that Gd accumulates in the extravascular space is also not frequently accounted for by this analysis. With Fermi-function deconvolution the central volume principle is used to describe the amount of Gd present within a region of myocardium.(23) An empirical Fermi-function is used for deconvolution and the initial amplitude of the Fermi-function fit is proportional to myocardial perfusion. Figure 1 demonstrates the quantification of perfusion using Fermi-function deconvolution of data acquired with a dual-bolus sequence. This technique is relatively robust to the effects of extracellular accumulation of the contrast agent at least during first pass of the contrast.(24) Multiple other “model-independent techniques” have been used for deconvolution, and recently there have been studies comparing the relative merits of the different techniques.(25) Both of the techniques described above have been shown to correlate with myocardial perfusion over a wide range of flow.

Figure 1. Absolute Quantification of Perfusion using Fermi-function deconvolution from a dual-sequence spiral trajectory acquisition.

(a) The myocardium is segmented to generate time intensity curves for the myocardial tissue function (TF). (b) The signal intensity is converted into concentration of gadolinium for both the arterial input function (AIF) and tissue function prior to deconvolution. (c) a pixel-by-pixel perfusion map shows uniform perfusion of 1mL/g-min in this normal volunteer.

Arterial spin labeling (ASL) is an alternative technique for quantitative perfusion which uses the water in blood as an endogenous contrast agent. The main limitation of ASL is poor SNR due to the small difference between the “labeled” and non-labeled acquisitions. This technique has been used to detect obstructive CAD; however there are still some technical hurdles for the technique to have widespread clinical applicability.(26) ASL may have utility in quantifying perfusion in more diffuse processes such as microvascular dysfunction which may not require high spatial resolution.

Clinical Quantitative Perfusion Imaging

Multiple studies have validated CMR perfusion imaging in experimental animal models with microspheres as the gold standard. CMR has also been validated in human subjects by direct comparison to PET and invasive measures of coronary flow reserve (CFR) and fractional flow reserve (FFR). A study of 19 healthy volunteers demonstrated strong correlation of both absolute perfusion (R=0.86) and MPR (R=0.96) between quantitative CMR using kinetic modeling and 13N-amonia PET. However there was an underestimate of peak perfusion by CMR as compared to PET.(27)

Quantitative MPR was compared to CFR as assessed by coronary flow wire in a study of 20 patients with suspected CAD. This study demonstrated a strong correlation between CMR and CFR (R=0.86) and a sensitivity of 88% and a specificity of 90% for predicting a CFR <2.(28) Quantitative CMR was also prospectively compared to FFR in 37 patients with suspected CAD who underwent FFR and coronary angiography.(29) The study showed a reduced MPR of 1.54 for segments with FFR<0.75 and an MPR of 2.11 for segments with FFR>0.75. A cutoff of 2.04 for MPR was highly sensitive (92.9%), but not specific (56.7%) for predicting an FFR<0.75. Our group studied 41 subjects with known or suspected CAD using quantitative MPR by CMR as compared to x-ray angiography for detection of CAD and delineation of the extent of ischemic myocardium (figure 2).(2) While qualitative and quantitative analysis had similar diagnostic accuracy (83% versus 80%) for detecting significant CAD, MPR detected a significantly higher burden of ischemia in three vessel disease as compared to single vessel disease (60% vs 25% p=0.02). For visual analysis, there was no difference in the extent of detected ischemia between three vessel and single vessel disease (31% vs 21% p=0.26). A recent study has directly compared the performance of quantitative CMR at 1.5T and 3T field strengths to detect an FFR<0.8 in 34 patients with known or suspected CAD. This study demonstrated a higher ROC area under the curve, (0.963 vs. 0.645, p < 0.001), sensitivity (90.5 vs. 61.9 %) and specificity (100 vs. 76.9 %) at 3T vs. 1.5T.(30)

Figure 2. Quantitative Perfusion may be useful in determining extent of ischemia in multi-vessel disease.

(a) Visual analysis of the stress images does not show a significant perfusion abnormality, however quantification shows a region of reduced perfusion reserve corresponding to an abnormality found in the LAD at cardiac catheterization. (b) Although visual and quantitative analysis had similar diagnostic accuracy, quantitative analysis revealed a larger area of ischemia in 3 vessel disease as compared to single vessel disease and may be important for quantifying the volume of ischemic myocardium. (Adapted from Patel et al. figures 4 and 5(2))

Current limitations and Future Potential

Several factors have hampered the widespread clinical application of quantitative perfusion CMR. One is the limited spatial coverage (typically 3–4 slices) as compared to PET. However new imaging pulse sequences using spatio-temporal parallel techniques have demonstrated the potential for rapid high resolution and 3D coverage of the myocardium.(18) Furthermore spiral or radial techniques hold promise for further acceleration of image acquisition.(31,32) Increased field strength from 1.5T to 3T may also improve SNR and diagnostic performance. The lack of robust automated tools for image registration and quantification has limited the applicability of quantitative analysis to research settings. However, non-rigid registration techniques have recently been applied to perform point-by-point quantification (Figure 3).(33) The final barrier is to demonstrate that perfusion quantification has additional benefit over visual analysis or semi-quantitative techniques. One of the justifications for absolute quantification has been improved detection of 3 vessel disease; however, the improved spatial resolution of CMR may provide sufficient gradient of perfusion to enable adequate visual analysis. Multiple studies have demonstrated a high diagnostic accuracy of CMR stress imaging without the complex data analysis required for quantification, and the incremental benefit of absolute quantification still needs to be established in larger clinical studies.(34)

Figure 3. Pixel-wise CMR perfusion maps at stress and rest from a healthy volunteer (1) and from 3 patients with coronary artery disease.

High quality pixel-wise maps are feasible using non-rigid registration for motion correction. (Adapted from Hsu et. al. figure 7 (33))

Quantification of Myocardial T2* Relaxation Times

Quantification of T₂* relaxation times is arguably the most clinically established quantitative CMR tissue characterization technique as it has revolutionized the detection and monitoring of iron-overload cardiomyopathy. T2* is the transverse relaxation time in the presence of static magnetic field inhomogeneities. T₂* mapping techniques enable detection of spatially varying differences in T₂*relaxation times and can be used to detect the presence of hemorrhage in acute myocardial infarction (MI). T₂* mapping techniques also can detect changes in myocardial oxygenation based on the blood oxygen dependent (BOLD) effect which results from the difference in magnetic state between oxyhemoglobin (diamagnetic) and deoxyhemoglobin (paramagnetic) as deoxyhemoglobin reduces the T₂* of blood.

CMR techniques for Quantifying Myocardial T₂*

To create a myocardial T2* map, multiple images with different sensitivity to T₂* must be acquired. This is typically achieved by collecting gradient echo images with different echo times (TE) yielding images with signal intensities which follow a T₂* relaxation curve. The most commonly used pulse sequence for data acquisition is a multi-echo segmented gradient echo technique which typically collects data at 8 different echo times ranging from 2ms to 20ms (Figure 4).(35) Image data is acquired directly after the R-wave which further reduces artifacts from blood flow and myocardial motion. T₂* measurements have primarily been measured in the intraventricular septum as this area is typically free from magnetic susceptibility artifacts which corrupt the long TE images, particularly in the lateral wall.

Figure 4. Calculation of T2* maps in a normal subject.

(a) A set of gradient echo images acquired with different echo times (TE) from 2ms to 18 ms are acquired. As the TE is increased the signal intensity decreases due to static field inhomogeneities resulting in T2* decay. (b) The data from each pixel is fit to a T2* decay curve. Pixels with longer T2* decay more slowly (red curve) as compared to regions with shorter T2* blue curve). (c) the T2* map shows a region of reduced T2* in the inferolateral wall which is caused by susceptibility artifact and can be seen even in normal subjects.

Clinical Application of T2* mapping techniques

The T2* relaxation time is shortened by magnetic field inhomogeneities induced by myocardial iron deposition. A reduction of T2* to less than 20 ms is consistent with the diagnosis of iron overload cardiomyopathy.(8) The presence of abnormal T2* is the most important predictor of future requirement for chelation therapy. Changes in the myocardial T2* has been used as a surrogate endpoint of several clinical trials of chelation therapy in transfusion dependent thalassaemia major patients.(9) Quantitative T2* mapping has been applied to the detection of myocardial hemorrhage in acute MI. In a study of 62 patients with acute reperfused MI, 62% of subjects had evidence of microvascular obstruction seen on late gadolinium enhanced (LGE) imaging, however only 20% of patient had hemorrhage as indicated by T2*<30 ms.(36) T2* mapping has been used to probe the BOLD effect in ischemic heart disease. An early study of T2* mapping in 16 healthy volunteers and 16 patients with known single vessel CAD demonstrated that volunteers had a T2* of 35ms at rest which increased to 40ms with dipyridamole infusion due to the reduction in deoxygenated hemoglobin during hyperemia. In subjects with single vessel CAD, there were regional reductions in T2* at rest in the regions supplied by stenotic arteries, and these differences increased significantly during dipyridamole.(37) In another study, 46 patients with known or suspected CAD underwent BOLD T2* mapping at rest and during adenosine stress at 3T. BOLD CMR at rest demonstrated significantly lower T2* values for ischemic segments (26.7± 11.6 ms) as compared to normal (31.9± 11.9 ms) segments. A stress T2* less than 33.8 had 78% sensitivity and 68% specificity on a per patient basis to detect significant CAD.(38)

Potential Clinical Pitfalls and Solutions

Care must be taken when interpreting regional differences in T2* from maps as the T2* can be abnormally short in certain regions of the heart even in normal subjects due to macroscopic magnetic field inhomogeneity. This is particularly problematic in the inferolateral wall. For assessment of iron overload which diffusely affects the myocardium, measurement of the T2* time in the intraventricular septum is preferred.

Quantification of Myocardial T₂ Relaxation Times

It has been long recognized that T₂ relaxation times are sensitive to myocardial edema and are known to be elevated in acute MI(39), myocarditis(40), sarcoidosis(41) and in cardiac allograft rejection(42). T₂ is the relaxation time for transverse magnetization and is prolonged in regions of edema or inflammation. The current standard T2-weighted dark-blood turbo spin echo sequences (DB-TSE) have significant limitations, particularly for semi-quantitative evaluation. Factors such as incomplete blood suppression with dark-blood techniques, regional variations in signal intensity from coil inhomogeneities, and signal loss from myocardial contraction and relaxation during acquisition are significant challenges for this technique in clinical practice. In addition, as T₂-weighted images are inherently non-quantitative, semi-quantitative techniques have relied on defining abnormal T₂ by a number of standard deviations of signal intensity over that of remote myocardium which can be problematic when there is no remote “normal” reference region. In the last few years, techniques capable of rapid measurement of the myocardial T₂ relaxation time have emerged and may have important advantages over the prior semi-quantitative techniques.

CMR techniques for Quantifying T2 relaxation times

In order to quantify myocardial T₂ relaxation times, multiple images with different sensitivity to T₂ need to be acquired. This is typically achieved by collecting spin-echo images with different echo times yielding images with signal intensities that follow a T₂ decay curve (figure 5). For cardiac T₂ mapping applications two different types of sequences have been employed: (1) Dark-blood turbo-spin echo based pulse sequences (DB-TSE) and (2) Bright-Blood T₂ magnetization-preparation (T2-Prep) based pulse sequences. Both free-breathing navigator-gated acquisition and breath-hold strategies have been utilized.

Figure 5. Calculation of T2 maps using a T2-PREP pulse sequence in a patient with atypical Takusubo cardiomyopathy.

(a) Multiple images with T2-Prepations with different TE are acquired. As the TE is increased for this spin-echo-based preparation, the myocardial signal intensity decreases due to T2 decay. (b) The data from each pixel is fit to a T2 decay curve, pixels with longer T2 (red curve) decay more slowly than regions with shorter T2 (blue curve). (c) The T2 map shows a region of edema (Yellow arrows and Red ROI) in this patient. (d) The absence of LGE confirms the diagnosis of atypical Takusubo cardiomyopathy.

TSE-based T₂-mapping techniques collect images at multiple echo times providing multiple points along the T₂ relaxation curve for data fitting. TSE pulse sequences are sensitive to ghosting artifacts due to blood flow and dark-blood preparation must be utilized with these strategies. Furthermore, these sequences have some inherent sensitivity to motion which needs to be accounted for to accurately quantify T₂. The effects of coil inhomogeneity are eliminated during the T₂ fitting procedure. The T2-prep pulse sequence consists of a spin-echo like-preparation module to impart the T2-weighted contrast and a readout scheme to rapidly collect the image data.(43) Typically a rapid steady-state free precession (SSFP) readout is used, which is less sensitive to some of the artifacts which limit TSE-based techniques. By acquiring multiple images, each with a different T2-prep duration, a T₂ map can be determined (Figure 5). (44,45) These techniques demonstrate an approximately 10% overestimate of T₂ relative to reference techniques, likely due to the SSFP readout which induces mixed T₁ and T₂ contrast.(44)

Clinical Application of T2 mapping Techniques

T2 mapping techniques have demonstrated utility for evaluation of a variety of cardiac pathologies. The edematous territory measured by T₂-weighted imaging and T₂ mapping has been shown to correlate with the region of ischemic injury in a canine model of MI.(46,47) This has stimulated considerable clinical interest in using T₂-weighted imaging to assess myocardial salvage.(48) A T₂-Prep SSFP based mapping technique was used to evaluate 27 patients with acute MI and 21 normal volunteers, demonstrating that the average T₂ in edematous myocardium was 69ms compared to 56ms in remote (Figure 6) and 55 ms in normal volunteers. (6) Importantly, edema was detected in 26 of 27 patients with T₂ mapping, whereas conventional STIR images were negative in 7 patients and uninterpretable in 2 cases.

Figure 6.

(a) T2 maps, T2-weighted and LGE images from patients with acute myocardial infarction. (b) The T2 relaxation time was significantly increased in the region of the infarct as compared to remote and healthy myocardium. (Adapted from Verhaert et. al. figures 1 and 2 (6))

In a study of 30 patients, T₂ mapping techniques have also demonstrated increased T2 in the involved segments of myocardium in patients with acute inflammatory diseases including myocarditis and Takotsubo cardiomyopathy (Figure 7).(7) In both pathologies there was no difference in the T₂ of the remote myocardium as compared to controls. A T₂ cutoff of 59 ms had a sensitivity of 94% and specificity of 97% for identifying affected myocardium. Conventional T₂-weighted images were uninterpretable in 7 subjects because of artifacts and did not demonstrate abnormalities in 2 subjects with abnormal T₂ values by T₂ mapping. Bright blood T₂-prep SSFP images did not show a clear region of increased signal intensity in 13 of 30 subjects. This study demonstrates that T₂ mapping may improve the ability to differentiate edematous myocardium over T2-weighted techniques.

Figure 7. T2 maps, T2-weighted and LGE images from 3 patients with acute myocarditis.

T2 is increased in multiple regions of the myocardium and may show more diffuse involvement than is evident on T2-weighted images. (Adapted from Thavendiranathan et. al. figure #(7))

T₂ mapping has also been used for the assessment of iron-overload cardiomyopathy.(49) The superparamagnetic properties of iron result in a shortening of both T₂ and T₂* relaxation times resulting in more rapidly decaying transverse magnetization which can be detected by CMR. T₂ mapping has two potential advantages over T₂* mapping in that pulse sequences for determination of T2 are less sensitive to static magnetic field inhomogeneities and typically produce images of higher SNR. In a study of 136 patients with thalassaemia, a linear relationship has been demonstrated between T₂ and T₂* (R=0.89) in subjects with iron overload with similar diagnostic utility for detecting iron overload.(49)

Potential Clinical Pitfalls and Solutions

Care must be taken when comparing absolute T₂ relaxation times between data acquired with different techniques as biases in the absolute numbers may occur between different sequences or patient-related factors. For the T₂-Prep mapping sequence, in a tachycardiac patient, incomplete T1 relaxation between each image acquisition may introduce some bias to the measured T₂-weighted images. This effect can be minimized by increasing the number of heart beats between images (i.e. from 3 to 4 relaxation beats to enable more time for T₁ relaxation. TSE-based pulse sequences can also have some errors in accurate T₂ determination by introducing some T1 dependence to the values. An additional problem is that current mapping sequences have limited spatial resolution as compared to T₂-weighted images; however, this problem is likely to be overcome by improved pulse sequences and reconstruction techniques.

Quantification of T1 relaxation times and Volume of Distribution (Vd) of Gadolinium

LGE imaging has become the gold standard technique for imaging focal myocardial fibrosis in CAD and non-ischemic cardiomyopathies.(50,51) However cardiac pathologies characterized by diffuse myocardial fibrosis cannot be evaluated adequately by LGE as there are no reference regions of normal myocardium. Figure 8 shows LGE images and histology from MI and dilated cardiomyopathy. In MI, LGE images show focal enhancement corresponding to the dense focal fibrosis in the infarct, whereas in dilated cardiomyopathy LGE images may not show any enhancement despite the presence of significant interstitial fibrosis. T₁ mapping techniques following Gd injection have demonstrated potential for evaluation of diffuse myocardial fibrosis and myocardial infiltration. However, the T₁ of the myocardium is a function not only of the amount of fibrosis, but also of Gd dose, clearance rate, and time after injection. There has been growing interest in assessing the volume of distribution of Gd, or extracellular volume (ECV), as a noninvasive marker of fibrosis by performing T₁ mapping before and after either a continuous infusion or bolus injection of Gd. The volume of distribution of Gd has also been used as a noninvasive marker of cardiac amyloidosis.(52) Non-contrast T₁ mapping, also referred to as native-T1 mapping, has demonstrated potential for assessing myocardial edema in MI similar to work that has been done with T2 -weighted techniques.(47) Non-contrast T1 values are increased in hypertrophic and dilated cardiomyopathy reflecting fibrosis(53), and in cardiac amyloidosis reflecting the presence of infiltration by amyloid protein.(52,54,55) T1 mapping and Vd measurements may also have potential for better delineation of peri-infarct zones and for providing a more quantitative assessment of myocardial scarring.

Figure 8. Calculation of post contrast T1 maps using a modified MOLLI pulse sequence in a patient with myocarditis.

Following an inversion pulse, images are obtained in subsequent heart beats to obtain images at multiple different inversion times during the same phase of the cardiac cycle. As the inversion time increases the longitudinal magnetization increases due to T1 recovery. (b) The data are sorted by inversion time and fit to T1 relaxation curves. Pixels with inflammation have shorter T1-s post contrast and recover more rapidly (red curve) than regions of normal myocardium. (d) T1 mapping shows two epicardial regions with inflammation (yellow arrows and red ROI). (e) By inverting the pixel values an R1 map can be generated which has a contrast appearance similar to conventional LGE images.

CMR techniques for Quantifying T1 Relaxation Times

There have been a variety of techniques that have been used to quantify myocardial T₁ values.(3–5,56–58). The simplest but least efficient method is to collect a single image with a given sensitivity to T₁ on each of multiple separate breath-holds and fit the signal intensities to the T1 recovery curve as shown schematically in figure 9. This allows a T1 map to be determined during a specific phase of the cardiac cycle but is time consuming since it requires multiple breath-holds and may be subject to image misregistration between breath-holds. Two techniques are available which acquire all of the data for a T₁ map in a single breath-hold. The traditional Look-Locker technique collects image data continuously to create images at different time points following a T₁-weighted inversion preparation. A limitation of the traditional Look-Locker techniques is that the heart is in a different phase of the cardiac cycle on each image, precluding mapping of a specific slice as there may be significant through plane motion.(59) The Modified Look-Locker (MOLLI) pulse sequence is a modification to the Look-Locker sequence in which single-shot images are obtained in diastole with 11 different T₁ sensitivities over 17 heart beats. The data from these diastolic source images are fit to the T₁ recovery function to create a T₁ map (figure 9). The MOLLI technique is reproducible, and produces source images with high signal-to-noise ratio.(60,61) A limitation of the standard MOLLI pulse sequence is the requirement of 17 heartbeats for data acquisition which may result in a breath-hold that is too long for some patients. Several modifications have been proposed to reduce the breath-hold duration needed to collect the data and to reduce the heart-rate dependence.(62,63) Furthermore, robust non-rigid registration has been introduced which minimizes the effect of motion between the images collected on different heartbeats.(64) The majority of recently published T₁ mapping research has utilized one of the three techniques, or some variation thereof.

Figure 9. Diffuse fibrosis is not apparent on conventional LGE images. (a) LGE images from an ischemic cardiomyopathy demonstrate a focal region of LGE corresponding to prior myocardial infarction.

(b) This region of increased signal intensity corresponds to focal fibrosis in the infarct on H&E stained myocardium. (c) LGE images from a patient with dilated cardiomyopathy do not demonstrate any focal LGE, however (d) histological evaluation would demonstrate diffuse interstitial fibrosis this is not identifiable by LGE imaging.

CMR protocols for determining the volume of distribution of Gd

By measuring T₁ before and after either a bolus injection or equilibrium infusion of Gd, the apparent volume of the extracellular space (ECV) can be determined by calculating the volume of distribution of Gd.(3,5,58,65) For the continuous infusion method, Gd is given as a continuous low dose infusion until the T₁ in the myocardium and blood pool are constant. The Vd can then be directly determined from the T₁ values obtained pre-contrast and at equilibrium according to the following equation.(5,65)

$$\mathit{ECV}=(1-\mathit{Hct})\lambda =(1-\mathit{Hct})\frac{T_{1{\mathit{post}}_{\mathit{myo}}}-T_{1{\mathit{pre}}_{\mathit{myo}}}}{T_{1{\mathit{post}}_{\mathit{blood}}}-T_{1{\mathit{pre}}_{\mathit{blood}}}}$$

The bolus injection method relies on the assumption that the exchange between the blood pool and myocardium is fast with respect to Gd clearance from the blood such that there is equilibrium between the concentration of Gd in the blood pool and myocardium at each time point following Gd administration.(3,58) The equivalence of these two approaches has been demonstrated.(62,66)

Clinical Application of T1 mapping Techniques

The reported normal T1 relaxation times for the myocardium at 1.5 T range between 900 and 1100ms with some variation based on technique.(59–61,63,66,67) Pre-contrast measurement of T₁ should be increased in myocardial edema similar to T₂ due to the increase in free water content of the tissue. A study of 8 patients who underwent pre-contrast T₁ mapping demonstrated an 18% increase in T₁ in infarcted regions.(68) The area of increased T₁ pre-contrast had a larger spatial extent as compared to the region of LGE.(68) Pre-contrast T₁ measurements have been compared to T₂ measurements as an alternative for assessing the “area at risk” following acute ischemic injury and produce results similar to microsphere measurements.(47) Pre-contrast T₁ times are prolonged in both the myocardium and blood pool of subjects with amyloidosis as compared to normal volunteers.(52,55). A recent study of 53 patients with AL amyloidosis demonstrated a significant prolongation of non-contrast T1 time in cardiac AL amyloidosis patients (1,140±61ms) as compared to 36 normal subjects (958±20ms). A T₁ cutoff of 1020 yielded 92% accuracy for identifying patients with possible or definite cardiac involvement.(54) There was no significant difference in non-contrast T₁ values between the controls and 17 subjects with aortic stenosis (979±51ms). Native T₁ was also shown to be prolonged in cardiomyopathy in a study of 25 patients with known hypertrophic cardiomyopathy and 27 subjects with non-ischemic dilated cardiomyopathy as compared to 30 control subjects.(53)

T₁ mapping post-contrast has demonstrated potential for evaluation of MI. In the same aforementioned study of 8 subjects with MI who underwent T1 mapping pre-contrast, there was a 27% reduction in post Gd contrast T1, with a spatial extent which correlated well with LGE, demonstrating that T₁ mapping could be an alternative for quantifying MI.(68) An image of R₁, which is defined as 1/T₁, will have an appearance similar to conventional LGE images where regions of fibrosis or infiltration are bright (Figure 9c–d)

T1 mapping has demonstrated the potential for assessing diffuse fibrosis and infiltration. In a study of 25 patients with heart failure, the T₁ times in the myocardium after Gd contrast were reduced compared to 20 normal controls, and the post-contrast T₁ times were correlated with the histological severity of fibrosis in 9 subjects who underwent myocardial biopsy.(4) This study demonstrated that post-contrast T₁ mapping could potentially be used to evaluate diffuse fibrosis in cardiomyopathy. A study of 54 heart failure patients undergoing endomyocardial biopsy demonstrated a reduction in the 10-minute post contrast T₁ time in the heart failure subjects as compared to 13 controls. In heart failure, the post-contrast T₁ time was inversely correlated with the histological severity of fibrosis.(69) T₁ mapping post-contrast has also shown potential for the evaluation of cardiac amyloidosis.(70) In 29 patients with proven cardiac amyloidosis a T₁ difference of 23 ms between the subepicardium and subendocardium at two minutes after Gd administration predicted mortality in amyloidosis with 85% accuracy.(70) In a study of 5 patients with amyloidosis, there was also a smaller difference in T₁ relaxation times between the blood pool and myocardium 17.1±54.3ms vs 136.1 ± 18.4 ms at 20 minutes after contrast administration as compared to 8 controls.(52)

By measuring T₁ before and after either a bolus or equilibrium infusion of Gd, the partition coefficient and volume of distribution of Gd can be assessed, and they are proportional to the extracellular volume (ECV) (Figure 10). The ECV of normal myocardium has been reported to be in the range of 24–28%.(56,62,66,71) Assessment of Vd or the partition coefficient lambda has been used to assess diffuse fibrosis in different cardiac pathologies. In 9 patients with dilated cardiomyopathy, the volume of distribution of Gd was found to be significantly increased in patients with both familial and dilated cardiomyopathy.(3) Similar findings were demonstrated in congenital heart disease with myocardial dysfunction.(56) Validation of the correlation between the Vd and histological evidence of myocardial fibrosis has been performed in 18 subjects with aortic stenosis undergoing aortic valve replacement and in 18 patients with HCM undergoing surgical myectomy using a multi-breath-hold equilibrium infusion T1 technique.(5) Given the prognostic data available in HCM for the evaluation of focal fibrosis with LGE, it is likely that T1 mapping may provide additional disease in this disease as it reflects both focal and diffuse fibrosis.(72)

Figure 10.

(a) Examples illustrating excellent agreement between LGE and ECV in cases of focal abnormalities in myocardial ECV. Pre-contrast T1-maps (top row), post-contrast T1-maps (2nd row), late gadolinium enhancement (3rd row), and ECV maps (bottom row) for patients with: (a) chronic MI, (b) acute myocarditis, and (c) HCM. (Adapted from Kellman et al. figure 4 (71))

The volume of distribution and the partition coefficient of Gd are markedly increased in amyloid cardiomyopathy with expansion of the extracellular space due to amyloid protein deposition.(55,73) The degree of change in Vd is much larger than the changes in the individual T₁ measurements of the blood pool and myocardium. This may serve as a new marker for determining disease burden in amyloid cardiomyopathy.

A direct comparison of ECV determined at 1.5T and 3T in 31 patients demonstrated similar ECV measurements between these two field strengths. However, the investigators noted differences in ECV between the septum and lateral wall which may be related to susceptibility and will likely be more problematic at higher field strengths.(74)

Using the pre and post contrast T1-maps, one can generate parametric images of Vd which reflect the extracellular volume (ECV) (Figure 10), thus providing a new type of image contrast which may provide insight into diseases characterized by regional differences in fibrosis.(71) ECV mapping may also have important prognostic significance in different cardiac pathologies.(75)

The development of robust techniques for non-rigid registration has greatly improved the clinical applicability of T1 mapping, as T1 maps can be generated without the need for user interaction (64). Further developments in parallel and compressed sensing techniques will further improve the spatial and temporal resolution of parametric mapping.

Potential Clinical Pitfalls and Solutions

It is important to note that the MOLLI T1 mapping pulse sequences have some heart rate dependence which may bias measured T1 relaxation times particularly in patients with fast heart rates. As this bias is T1 dependent, the effect is worse for pre-contrast T1 times where the T1 relaxation times are on the order of 1–1.5 seconds. For the MOLLI pulse sequence this effect can be mitigated to an extent by increasing the number of relaxation heart beats; however, this directly increases the required breath-hold. Secondly, the specific parameters of the SSFP readout module can also introduce some T2 dependence to the measured T1 values. Thus, some caution is warranted when comparing T1 measurements acquired by different methodologies.

Conclusions

Quantitative techniques for myocardial perfusion and parametric mapping of magnetic relaxation times are extending the unique potential of CMR for characterization of cardiac structure and physiology. There is a growing body of evidence for the clinical utility of quantitative assessment of perfusion and relaxation times, although current techniques still have some important limitations. There are significant differences in the techniques used for both image acquisition and image analysis between MRI vendors and different sites. Standardization of acquisition and analysis will be important for the wider clinical applicability of these techniques. Improved scanner technology and reconstruction and image processing techniques are rapidly evolving which will further improve the available techniques for CMR parametric mapping. Further research will be necessary to demonstrate that these new techniques will offer incremental diagnostic and prognostic utility and robustness as compared to current visual and semi-quantitative CMR techniques.