General review of magnetic resonance elastography

Abstract

Magnetic resonance elastography (MRE) is an innovative imaging technique for the non-invasive quantification of the biomechanical properties of soft tissues via the direct visualization of propagating shear waves in vivo using a modified phase-contrast magnetic resonance imaging (MRI) sequence. Fundamentally, MRE employs the same physical property that physicians utilize when performing manual palpation - that healthy and diseased tissues can be differentiated on the basis of widely differing mechanical stiffness. By performing “virtual palpation”, MRE is able to provide information that is beyond the capabilities of conventional morphologic imaging modalities. In an era of increasing adoption of multi-parametric imaging approaches for solving complex problems, MRE can be seamlessly incorporated into a standard MRI examination to provide a rapid, reliable and comprehensive imaging evaluation at a single patient appointment. Originally described by the Mayo Clinic in 1995, the technique represents the most accurate non-invasive method for the detection and staging of liver fibrosis and is currently performed in more than 100 centers worldwide. In this general review, the mechanical properties of soft tissues, principles of MRE, clinical applications of MRE in the liver and beyond, and limitations and future directions of this discipline -are discussed. Selected diagrams and images are provided for illustration.

Key Words: Magnetic resonance elastography; Elasticity imaging techniques; Liver disease; Fibrosis; Emerging applications

Core tip: Magnetic resonance elastography (MRE) is a non-invasive technique for probing the mechanical properties of biologic tissues. The three main steps involved in the MRE technique include the production of mechanical waves in soft tissues, the adoption of a modified phase-contrast MR sequence to image wave motion, and the application of an inversion algorithm to convert the wave image into an elastogram. MRE has received validation for the non-invasive assessment and grading of fibrosis in chronic liver disease patients. MRE also has potential diagnostic applications in other organ systems and may help further the understanding of disease processes.

INTRODUCTION

Magnetic resonance elastography (MRE) is a rapidly emerging non-invasive diagnostic technique for the detection and characterization of a wide range of disease processes. By interrogating the biomechanical properties of tissues, which range widely in physiologic and pathologic states, MRE renders information that is beyond the capabilities of conventional imaging modalities that rely almost exclusively on morphological changes for detecting disease. In an era of increasing adoption of multi-modality techniques for evaluating complex disease processes, MRE may be incorporated into a standard magnetic resonance imaging (MRI) examination to provide a rapid, reliable and comprehensive imaging assessment at a single patient appointment. The largest amount of clinical experience in MRE is in the evaluation of chronic parenchymal liver diseases although MRE has applicability in other organs systems as well. In this general review - the mechanical properties of soft tissues, principles of MRE, clinical applications of MRE in the liver and beyond, and limitations and future directions of this discipline - are discussed.

MECHANICAL PROPERTIES OF SOFT TISSUES

Elasticity imaging is a field of medical science that involves the profiling of the mechanical properties of biologic tissues for the detection and characterization of disease. This discipline derives its motivation from palpation, a clinical assessment technique that has stood the test of time, having been passed down to successive generations of physicians for centuries. When tissues become pathologic, biomechanical changes occur which alter the rigidity of these tissues, a phenomenon perceived on palpation as increased stiffness. The term “stiffness”, a biomechanical property of tissue, represents its ability to resist deformation when subjected to a force. The mechanical properties of biologic tissues, including between physiologic and pathologic states, vary widely by more than 4 orders of magnitude[1,2]. In comparison, conventional modalities such as ultrasound (US), computed tomography and MRI vary over a much narrower diagnostic range. Since its clinical introduction 2 decades earlier, many elasticity imaging techniques have been explored. These applications vary in several ways including[1-6]; (1) the origin of the stress (external/internal); (2) the temporal characteristics of the stress (static/quasi-static/dynamic); and (3) the imaging platform (optical imaging/ultrasound elastography/MRE) used. Irrespective of the particular technique in question, all applications share the same foundational concept. This involves (1) generating a stress in a target tissue; (2) examining the tissue’s strain response; and (3) characterizing the stress-strain dynamics of the tissue. In material science, stress is defined as “force applied per unit area” while strain is defined as “tissue displacement per unit sample length”. The slope of the stress-strain curve is the “elastic modulus” - a physical parameter that reflects the intrinsic stiffness of the tissue (Figure 1).

Figure 1 Schematic diagram illustrates the stress-strain relationship of elastic materials and viscoelastic materials.

Purely elastic materials demonstrate “spring-like” behavior, where stress and strain are linearly related according to Hooke’s Law. However, most biologic tissues are viscoelastic and show both elastic and viscous properties. In these tissues, the elastic modulus is a complex entity that comprises a storage modulus (representing elasticity) and a loss modulus (representing viscosity). The viscosity component exhibits a “dashpot like” behavior (damping), where the strain rate varies with time and is characterized by a hysteresis loop due energy losses from loading and unloading of stresses. Mathematical models such as described by Maxwell, Volgt, and Kelvin (using different combinations of springs and dashpots) have been developed to predict the stress-strain dynamics of viscoelastic materials[3]. Depending on the type of stress encountered, the elastic modulus is reported as Young’s modulus, E {[longitudinal (compressive or tensile) stress]/longitudinal strain}, shear modulus, μ {[shear (transverse) stress]/shear strain} or bulk modulus, K (volumetric stress/volumetric strain). The SI unit for elastic modulus is expressed in kilopascals (kPa) where 1 kPa = 1 kN/m². The relationship between E, μ and K is defined mathematically as:

μ = E/[2(1 + v)] and K = E/[3(1 - 2v)]

Since the Poisson’s ratio, v (tranverse contraction per unit breath/longitudinal extension per unit length) for most soft tissues is very close to an incompressible liquid (v = 0.5), E approximately equal 3μ. K does not vary significantly in biologic tissues[7]. MRE utilizes shear wave propagation within target tissues to calculate the shear modulus[8]. Assuming that biologic tissues are homogeneous, isotropic and linearly elastic, μ is calculated as follows[2]:

μ = ρv_s² or ρ벃² and v_s = λƒ

where v_s is the speed of the shear wave (m/s), λ is the wavelength of the shear wave (m), ƒ is the frequency of the shear wave (Hz) and ρ is the soft tissue density (assumed to be 1000 kg/m³)[2]. Given that these assumptions may be inaccurate as biologic tissues have viscoelastic properties, the term “shear stiffness” is used in MRE as an approximation for the effective shear modulus estimated at a specified frequency. Given the contribution of viscoelasticity on tissue stiffness, using the same excitation frequency is a key issue in maintaining consistency between MRE examinations. MRE may be performed using either a quasi-static or dynamic excitation - while the former offers a qualitative estimation of tissue stiffness, the latter is preferred as it allows a precise quantitative assessment of this parameter[8-11].

PRINCIPLES OF MRE

Dynamic MRE is a diagnostic technique that was originally described at the Mayo Clinic by Muthupillai and colleagues in 1995[8,12,13]. MRE can be performed on most standard MRI platforms, be this 1.5T or 3T, with the introduction of additional hardware and software[2,12] (Figures 2-4). The three main steps involved in the MRE technique include the production of mechanical waves in soft tissues, the adoption of a modified phase-contrast MR sequence to image wave motion, and the application of an inversion algorithm to convert the wave image into an elastogram[8,12].

Figure 2 Schematic diagram demonstrates the set up for an magnetic resonance elastography examination using a pneumatic actuator system. Reproduced with permission from “John Wiley and Sons”, Venkatesh et al[12].

Figure 3 Typical transcostal position of the passive driver for magnetic resonance elastography assessment of the liver. Reproduced with permission from “John Wiley and Sons”, Venkatesh et al[12].

Figure 4 Gradient-echo magnetic resonance imaging sequence with motion sensitizing gradient applied along the slice selection direction (G_Z) to detect cyclic motion in that direction. A phase offset (θ) between the MSG and the acoustic driver was adjusted to acquire wave images at different time intervals during a single period of wave motion. Reproduced with permission from “Elsevier”, Yin et al[13]. MSG: Motion sensitizing gradient.

Mechanical excitations are produced by actuator devices such as pneumatic, electromechanical and piezoelectric systems, which are positioned against the body surface close to the target organ. These devices generate longitudinal compression waves at skin surfaces, which are then mode converted at internal tissue interfaces into transverse shear waves. A pneumatic system incorporates a non-MRI compatible active driver (placed outside the scan room) for producing continuous acoustic wave motion, an MRI compatible passive driver placed against the body surface, and an air-filled plastic tube for transmitting the pneumatic excitations from the active to the passive driver. The active driver generally consists of a signal generator connected to an audio amplifier and a loudspeaker. The passive driver for most abdominal applications has a drum or disc-shaped design to maximize surface area contact, and this may be applied to the body in any orientation. It is often secured to the patient by an elastic strap. A transcostal position is favored over a subcostal position when imaging the liver, as trans costal excitation improves penetration as the ribs act as a secondary wave source[14]. Limitations of pneumatic actuators include susceptibility to phase delays that may affect synchronization with the MRI sequence, and reduced effectiveness at frequencies > 300 Hz[15,16]. Electromechanical actuators have the advantage of good synchronization with the MRI sequence and can produce high amplitude mechanical excitations at low power requirements[15]. However, these systems can cause electromagnetic interference with eddy current induction leading to MRI artifacts and heat build-up. As such, electromechanical actuators have to be positioned remote from the imaging region and have to conform to a fixed orientation with respect to the static magnetic field[15]. Piezoelectric actuators provide stable and precise mechanical excitation up to frequencies of 500 Hz but are relatively expensive[15].

The propagating shear waves are imaged using a phase contrast MRI technique that includes oscillating motion sensitizing gradients (MSGs). This may be achieved with a wide range of sequences such as gradient-echo, spin-echo, balanced steady-state free precession or echo-planar imaging. The MSGs measure tissue motion along a specified direction, and can be applied along all 3 orthogonal planes of motion to capture the entire 3-dimensional (3D) wave vector. Trigger pulses are used to synchronize the mechanical excitations with the MRI sequence, and the actuator and MSGs are typically set to the same frequency (e.g., 40-200 Hz)[17]. Shear wave motion in tissues induces cyclic spin displacement of protons, which in the presence of synchronized MSGs are encoded as phase shifts within the MRI signal. The phase shifts are calculated as follows[8]:

Math 1

Math 1 Math(A1).

Where φ is the phase shift, vector r is the spin position vector, Θ is the relative phase of mechanical and magnetic oscillations, ϒ is the gyromagnetic ratio, N is the number of gradient cycles, T is the period of the gradient waveform, G₀ is the MSG vector, vector Ε₀ is the displacement amplitude vector, and vector k is the wave vector[8]. The phase shift is proportional to the dot product of the MSG and the displacement vectors[8]. This information is used by the MRI signal to create phase and magnitude images. A curl filter is then applied to separate the shear wave data from the longitudinal wave data. The resultant MRI image, called a wave image, is a spin displacement map - a snapshot of the shear wave motion in the tissues (with sensitivity to displacements as small as 100 nm)[8]. Several wave images at different time points are acquired by varying the phase offset between the actuator and the MSGs. Typically, 4 to 8 phase offsets are applied.

Post-processing inversion algorithms, such as local frequency estimation and algebraic inversion of the differential equation, are used to transform the wave images into elastograms. The elastograms are quantitative maps of tissue stiffness and are displayed as both greyscale and color images. Tissue stiffness is assessed by performing region of interest tracings of the target organ, in areas that have adequate wave amplitude, and which do not extend closer than half a wavelength to the organ boundary[12]. Tracings should avoid large vessels and areas affected by motion artifact[12].

CLINICAL APPLICATIONS IN THE LIVER

The largest volume of collective experience on MRE is in the investigation of chronic liver diseases (CLD) (Figure 5). Multiple studies have confirmed the value of MRE for the non-invasive evaluation of liver fibrosis[13,14,18-23]. A systemic review and meta-analysis by Singh et al[22] of 12 studies involving 697 patients with a less than one year interval between MRE and liver biopsy reported that MRE had high accuracy for the diagnosis of significant (F2) or advanced fibrosis (F3) and cirrhosis (F4) independent of etiology of CLD or body mass index (BMI). On receiver operating characteristic (ROC) analysis, the area under the ROC curve and 95%CI for MRE were 0.84 (0.76-0.92) for ≥ F1, 0.88 (0.84-0.91) for ≥ F2, 0.93 (0.9-0.95) for ≥ F3 and 0.92 (0.9-0.94) for F4. The study also reported a low failure rate of 4.3% for MRE. A meta-analysis by Su et al[23] of 13 studies involving 989 patients also showed a high diagnostic performance for MRE in the detection and staging of liver fibrosis. The authors found pooled sensitivities and specificities for MRE of 87% (95%CI: 84%-89%) and 92% (95%CI: 87%-96%) for F ≥ 1, 87% (95%CI: 84%-90%) and 92% (95%CI: 89%-95%) for F ≥ 2, 88% (95%CI: 85%-91%) and 91% (95%CI: 88%-93%) for F ≥ 3, and 91% (95%CI: 87%-94%) and 92% (95%CI: 89%-94%) for F4. The pooled area under the ROC curve was 0.95 for F ≥ 1, 0.97 for F ≥ 2, 0.97 for F ≥ 3, and 0.98 for F4. In a study of 50 patients with CLD and 35 healthy volunteers, Yin et al[13] demonstrated that liver stiffness on MRE exhibited a consistent and systematic increase with increasing fibrosis scores (Figure 6). Using a threshold of 2.93 kPa at 60 Hz, Yin et al[13] found that MRE had 98% sensitivity, 99% specificity and 97% negative predictive value for liver fibrosis. MRE also performs well in direct comparisons with other tests such as serum fibrotic markers (APRI - aspartate aminotransferase to platelet ratio index, FIBROSpectII), ultrasound elastography and contemporary MRI techniques (e.g., dynamic contrast-enhanced MRI including gadoxetic acid enhanced MRI, diffusion-weighted imaging, etc.)[19,20,24-27]. In a retrospective study of 58 patients with non-alcoholic fatty liver disease (NAFLD), Chen et al[28] found elevated liver stiffness in patients with non-alcoholic steatohepatitis (NASH), where chronic inflammation was present without fibrosis. MRE was able to stratify NALFD cases into 3 categories - simple steatosis, NASH or fibrosis[28] based on the liver stiffness. Using a threshold of 2.74 kPa, MRE showed high accuracy for differentiating patients with NASH from those with simple steatosis with an area under the ROC curve of 0.93, sensitivity of 94% and specificity of 73%. The authors concluded that MRE had the potential to identify patients with steatohepatitis prior to the onset of fibrosis. Loomba et al[29] evaluated the diagnostic accuracy of MRE for the detection of advanced fibrosis (F3-4) in a prospective study of 117 consecutive patients with biopsy-proven NAFLD. The mean BMI was 32.4 ± 5 kg/m² and the median time from biopsy to MRE was 45 d. Using a threshold of ≥ 3.63 kPa, MRE had an area under the ROC curve of 0.92 for differentiating F3-4 from F0-2 with 86% sensitivity (95%CI: 65%-97%), 91% specificity (95%CI: 83%-96%), 68% positive predictive value (95%CI: 48%-84%) and 97% negative predictive value (95%CI: 91%-99%). In a randomized double-blind trial by Loomba et al[30] of 50 patients with biopsy-proven NASH, ezetimibe (a drug that inhibits intestinal fat absorption) was compared with placebo - treatment efficacy was assessed using a combination of MRE, MRI derived proton density fat fraction (PDFF) and liver biopsy. Although the study found that ezetimibe did not significantly reduce hepatic steatosis, the authors deemed that the trial was successful in demonstrating that MRE combined with PDFF was a feasible non-invasive method for evaluating treatment response in NASH.

Figure 5 Largest volume of collective experience on magnetic resonance elastography is in the investigation of chronic liver disease. A: A 32-year-old female with PSC disease stage 1 with normal liver stiffness; B: A 53-year-old female with hepatitis C and presumed cirrhosis, which was determined by MRE results. PSC: Primary sclerosing cholangitis; MRE: Magnetic resonance elastography.

Figure 6 Liver stiffness on magnetic resonance elastography increases systematically with greater fibrosis extent as determined by liver biopsy. Reproduced with permission from “Elsevier” Yin et al[13].

Studies have confirmed that liver MRE is a reliable test with good repeatability and inter-rater agreement[31-34]. While most investigations were performed on a 1.5T MR system, Shi et al[35] showed in 22 healthy volunteers that liver MRE performed on a 3T MR system had good short and mid-term (within 6 mo) repeatability. In a study of 41 healthy Asian volunteers, Venkatesh et al[36] demonstrated that liver stiffness measurements on MRE were not significantly influenced by age, gender, BMI or hepatic steatosis. For liver MRE examinations performed 4 to 6 wk apart, the study found an intraclass correlation coefficient (ICC) of 0.9 (95%CI: 0.78-0.96) and a within-subject coefficient of variation of 2.2%-11.4%. Serai et al[37] found consistent liver stiffness measurements (ICC = 0.994, P < 0.01) in 13 participants (8 healthy volunteers and 5 patients) examined on the same day using 2 separate 1.5-T MR vendor platforms (Phillips and General Electric). On Bland-Altman analysis, the mean stiffness difference between vendor platforms was 0.09 kPa and the 95% limits of agreement were 0.34 kPa and -0.16 kPa. Hallinan et al[38] performed liver MRE twice (before and 5 min after intravenous gadolinium administration) in 210 patients with CLD on the same 1.5T MR system. The study found that gadolinium did not significantly affect the diagnostic performance of MRE for detecting significant liver fibrosis. With the advent of multi-parametric liver imaging, the flexibility of being able to perform MRE either before or after gadolinium administration is an advantage. To determine if participants require to be fasted for liver MRE, a few studies have evaluated whether liver stiffness measurements differ when performed before a meal vs after a meal[33,39,40]. In 12 healthy volunteers, Hines et al[33] found that liver MRE had a within-subject standard deviation of 8.5% and 9% for fasted and fed states, respectively. Yin et al[39] performed liver MRE before and after a meal in 25 patients with biopsy-proven fibrosis and 20 healthy volunteers. Patients with fibrosis showed a significant post-prandial liver stiffness increase of 0.89 ± 0.96 kPa or 21.24% ± 14.98%. Healthy volunteers showed a smaller but significant post-prandial stiffness increase of 0.16 ± 0.20 kPa or 8.08% ± 10.33%. In a study of 19 CLD patients and 11 healthy volunteers, Jajamovich et al[40] showed that liver stiffness measurements on a 3T MR system were significantly higher for the post-prandial state compared with the pre-prandial state. The liver stiffness measurements were also significantly higher in patients compared with healthy volunteers. The mean liver stiffness for CLD patients was 4.9 ± 1.4 kPa for a fasted state vs 5 ± 1.2 kPa for a fed state (P < 0.001) while in healthy volunteers it was 1.8 ± 0.2 kPa for a fasted state and 2 ± 0.2 kPa for a fed state (P < 0.001). The authors concluded that liver MRE should be performed in a fasted state.

MRE may have a role as an alternative to liver biopsy for well-selected patients[12,27]. Biopsy is invasive and is associated with poor patient acceptability and risk of complications. It is also prone to sampling errors and may be non-diagnostic if the sample volume is insufficient or of poor quality. Given its non-invasiveness, MRE is an attractive option for longitudinal patient monitoring, including evaluating treatment efficacy. If a liver biopsy is necessary, MRE could be used to target biopsy to areas where the fibrosis is most severe[41]. A preliminary study involving 44 liver tumors found that MRE was able to differentiate tumors as malignant or benign with 100% accuracy, using a stiffness threshold of 5 kPa at 60 Hz[42]. A recent study of 79 patients with 80 malignant and 44 benign focal liver lesions found that MRE was superior to diffusion weighted imaging (DWI) for differentiating between malignant and benign etiologies[43]. Malignant lesions exhibited a significantly higher mean stiffness (7.9 kPa vs 3.1 kPa, P < 0.001) and lower mean apparent diffusion coefficient value (129 vs 200 × 10^-3 mm²/s, P < 0.001) compared with benign lesions[43]. ROC analysis found a significantly higher accuracy for MRE than DWI (0.986 vs 0.82, P = 0.0016)[43]. The study also showed higher stiffness in cholangiocarcinomas compared with hepatocellular carcinomas and focal nodular hyperplasias compared with adenomas.

US elastography methods such as transient elastography (TE) and shear wave elastography (SWE) are alternative techniques to MRE. Each modality has its own merits and limitations. MRE offers wide field imaging coverage and can also be performed as a 3D technique. In contrast, TE is a 1D technique and is limited to narrow field imaging to a maximum depth of 8 cm. As such, MRE has superior diagnostic accuracy compared to TE and provides a more representative delineation of tissue stiffness[20,44]. This is particularly relevant in liver fibrosis which may affect the liver heterogeneously. In a study of 113 patients with CLD, Ichikawa et al[44] found an area under the ROC curve of 0.97 (MRE) vs 0.93 (TE) for F4 (P = 0.03), 0.98 (MRE) vs 0.87 (TE) for ≥ F2 (P = 0.0003) and 0.97 (MRE) vs 0.87 (TE) for ≥ F1 (P = 0.01). In a study of 129 patients, Yoon et al[45] showed that liver stiffness measurements were more reliable on MRE compared with SWE although both techniques showed moderate correlation (r = 0.724) and comparable diagnostic performance for ≥ F2. In a study of 141 patients with CLD, Huwart et al[20] found that MRE had a greater technical success rate than TE (94% vs 84%) - unlike the latter, MRE may be performed in obese subjects and in those with ascites. The merits of US elastography compared with MRE include (1) lower equipment costs; (2) more established technique; (3) shorter examination length; (4) portability (e.g., can be performed at the bedside in unstable patients); and (5) may be performed in claustrophobic patients and those with MRI contraindications. In practice, both modalities are often viewed as complementary, where one technique may be preferred over the other depending on specific circumstances.

CLINICAL APPLICATIONS BEYOND THE LIVER

Several preliminary studies have assessed the utility of MRE in other organs. However, more evidence is needed before definitive recommendations can be made.

Spleen

A couple of studies have evaluated MRE in the spleen (Figure 7). A study of 16 healthy volunteers by Mannelli et al[46] found that splenic stiffness (mean of 3.57 ± 0.59 kPa at 60 Hz) was not influenced by BMI, mean arterial pressure, age, splenic volume or liver stiffness. Talwalkar et al[47] found that splenic stiffness was significantly higher in patients (n = 38) with CLD compared with controls (n = 12), with a splenic stiffness ≥ 10.5 kPa at 60 Hz reported in patients with esophageal varices. In a study of 139 patients, Shin et al[48] found a significant positive linear correlation between splenic and hepatic stiffness measurements and the grade of esophageal varices. Furthermore, the authors found that splenic and hepatic stiffness measurements showed superior performance compared to splenic length in predicting high-risk varices. In 10 patients, Guo et al[49] found a linear correlation between splenic viscoelasticity on multi-frequency MRE and hepatic venous pressure gradient (HVPG) measurements (reflecting portal pressure) before and after transjugular intrahepatic portosystemic shunt placement. On multi-frequency MRE in 36 patients, Ronot et al[50] showed that the splenic loss modulus (representing viscosity) was the best parameter for the non-invasive evaluation of HVPG -this showed potential for detecting high-risk esophageal varices and severe portal hypertension.

Figure 7 A patient with non-alcoholic steatohepatitis has a cirrhotic liver with stigmata of portal hypertension including splenomegaly, ascites, esophageal varices, splenorenal shunts and other collaterals. The left image shows mechanical waves throughout the liver and the spleen. In the resulting elastogram (right), both the liver and the spleen show very high shear stiffness values of over 20 kPa.

Kidneys

A few studies have explored the utility of MRE in the kidneys (Figure 7). A study involving 16 healthy volunteers found that renal MRE was a reliable technique as assessed by test-retest repeatability and inter-rater agreement with ICC values > 0.9. The study also found a within-subject coefficient of variation of 7.7% at 60 Hz and 13.6% at 90 Hz[51]. Rouvière et al[52] also reported that renal MRE was a reliable technique with a mean variation of 6% (2%-16%) at 45 Hz and 6% (1%-14%) at 76 Hz in 10 healthy volunteers. Correlating MRE findings with renal transplant biopsy, Lee et al[53] found a non-significant trend towards higher stiffness in moderate fibrosis (n = 2) compared with mild fibrosis (n = 6)[53]. Interestingly, the renal stiffness was slightly higher in a single case of no fibrosis at 120 and 150 Hz, and slightly lower at 90 Hz, compared with 2 cases of moderate fibrosis. The findings should be interpreted with caution given the small sample size. A study of 21 patients with cirrhosis and ascites found that renal MRE has potential for the evaluation of hepatorenal syndrome (HRS)[54]. The study found that renal stiffness in patients with HRS (median stiffness of 3.30 kPa and 2.62 kPa at 90 Hz and 60 Hz, respectively) was significantly lower compared with patients with normal renal function (median stiffness of 5.08 kPa and 3.41 kPa at 90 Hz and 60 Hz, respectively)[54]. On ROC analysis, renal MRE had an area under the ROC curve of 0.94 at 90 Hz and 0.89 at 60 Hz for the detection of HRS. Renal MRE also had an excellent inter-rater agreement with ICC values > 0.9. These early studies provide motivation for new lines of research in the kidneys.

Breast

Preliminary clinical studies involving small numbers of patients suggest that MRE has the potential to distinguish between malignant and benign breast tumors, based on the higher stiffness of malignancies[55-65]. In a breast MRE study involving 20 patients and 15 healthy volunteers, Lorenzen et al[55] found a median stiffness of 15.9 kPa for malignant breast tumors, 7 kPa for benign breast tumors and 2.5 kPa for normal breast parenchyma at 60 Hz. In a study of 57 patients with breast lesions, Siegmann et al[65] found that the addition of breast MRE to conventional contrast-enhanced MRI improved diagnostic yield for lesion characterization.

Brain

Imaging the mechanical properties of the brain may open up new possibilities for the study of brain function and neurologic disease (Figure 7). The MRE measured stiffness varies between different regions in the brain and between healthy and diseased states. In healthy volunteers, stiffness varies between the cerebrum and the cerebellum and between grey and white matter[66-69]. By probing the mechanical properties of brain tissue, MRE may contribute to the understanding of diseases such as multiple sclerosis, Alzheimer’s disease, traumatic brain injury and normal pressure hydrocephalus[70-76]. Preliminary investigations on brain tumors are encouraging and suggest that MRE offers fresh information beyond that of conventional modalities[77,78]. A study by Reiss-Zimmermann et al[79] in 27 patients found that MRE showed potential for characterizing intracranial neoplasms (e.g., glioblastomas, anaplastic astrocytomas, meningiomas and cerebral metastases) based on their mechanical properties including differentiating meningiomas from intra-axial neoplasms. Future work would include assessing if MRE can be used to guide surgical planning including evaluating the adequacy of resection (Figure 8).

Figure 8 A 2-dimensional multislice magnetic resonance elastography sequence was used to collect 3-dimensional vector wave images in a healthy volunteer at 90 Hz. The elastogram from the 3D direct inversion is shown. The mean shear stiffness of the right kidney is 5.8 + 0.34 kPa and the left kidney is 6.1 + 0.42 kPa. 3D: 3-dimensional.

Musculoskeletal

MRE shows promise for the biomechanical analysis of diseased and healthy skeletal muscle[80-89]. It has the potential to characterize neuromuscular diseases and to evaluate treatment efficacy[83,84]. In the assessment of muscle function, MRE shows good correlation with electromyography and is sensitive to age-related changes in muscle mechanical properties[85-87]. Both shear wavelength and tissue stiffness increase with greater muscle loading, while elevated muscle stiffness has been found in patients with neuromuscular disorders[88]. In a study involving 9 patients with active myositis, a significant reduction in muscle stiffness was observed during the relaxation phase in diseased muscles compared with controls[89]. The MRE technique for musculoskeletal assessment can be further extended to a 3T MR system (benefits include shorter examination lengths and superior signal to noise ratio). A study of 16 healthy volunteers by Hong et al[90] showed that MRE at 3T was a feasible technique for assessing shoulder muscle stiffness with good to excellent levels of inter-observer agreement (Figure 9).

Figure 9 A 39-year-old healthy male volunteer. Axial images of the brain including 60 Hz curl images and the 3D direct inversion elastogram. 3D: 3-dimensional.

Miscellaneous

Early studies suggest that MRE may have a role in evaluating cardiac function (e.g., left ventricular contractility), estimating chamber pressures (e.g., end-diastolic left ventricular pressure) and detecting abnormal ventricular relaxation[91-96]. A few investigators have performed MRE in the pancreas[97], uterus and cervix[98], prostate[99-106], the abdominal aorta[107-110], the lungs[111-115] and the head and neck[116,117]. These early studies should act as a stimulus for further research activity.

LIMITATIONS AND FUTURE DIRECTIONS

Hepatic MRE has limited utility in iron-overload states. On a gradient-echo MRE sequence, parenchymal iron-overload may result in the MRI signal being too low for shear wave detection. This problem can be mitigated by utilizing spin-echo or echo-planar imaging as these sequences are less susceptible to T2* effects. As in conventional MRI, MRE cannot be performed in patients with recognized MRI contraindications (e.g., cardiac pacemakers, cerebral aneurysm clips, cochlear implants, etc.) and may be impractical in patients with claustrophobia. There are also challenges involved in performing MRE in unstable patients given the longer examination times involved and in morbidly obese patients who may have problems fitting into the MRI scanner. It should be noted that the evaluation of parenchymal organ fibrosis may be limited by confounders of increased tissue stiffness. In the liver, alternative causes of elevated stiffness include acute inflammation (e.g., alcohol and viral hepatitis), biliary obstruction, passive hepatic congestion from cardiac failure, and hepatic venous obstruction. In the kidneys, alternative causes of elevated stiffness include renal vein thrombosis, hydronephrosis, renal compression from adjacent peri-renal collections, and possibly hypertension. While early studies suggest that MRE shows promise in differentiating benign from malignant tumors in a variety of organ systems, an overlap in stiffness is a limiting factor and prospective studies involving larger numbers of patients are required for validation.

Future directions involving MRE include (1) improving image quality and reducing acquisition time (e.g., use of echo-planar imaging, parallel imaging, reduced k-space acquisitions, imaging at ≥ 3 T, adoption of sophisticated inversion algorithms, 3D- and multi-frequency techniques, etc.)[118-124]; (2) expanding research into topics beyond fibrosis (e.g., inflammation, necrosis, edema, perfusion, tumor characterization, evaluation of treatment response, etc.)[12,28,125-128]; (3) better defining the pitfalls of MRE and developing novel solutions for overcoming these limitations; and (4) better defining the clinical indications for MRE beyond the research setting. Unquestionably, MRE is an emerging technology with genuine promise but the technique has yet to be fully mapped out and new frontiers exist for exploration.