Clinical application of magnetic resonance elastography in pediatric neurological disorders

Abstract

Magnetic resonance elastography is a relatively new, rapidly evolving quantitative magnetic resonance imaging technique which can be used for mapping the viscoelastic mechanical properties of soft tissues. MR elastography measurements are akin to manual palpation but with the advantages of both being quantitative and being useful for regions which are not available for palpation, such as the human brain. MR elastography is noninvasive, well tolerated, and complements standard radiological and histopathological studies by providing in vivo measurements that reflect tissue microstructural integrity. While brain MR elastography studies in adults are becoming frequent, published studies on the utility of MR elastography in children are sparse. In this review, we have summarized the major scientific principles and recent clinical applications of brain MR elastography in diagnostic neuroscience and discuss avenues for impact in assessing the pediatric brain.

Introduction

The basic principle of elastography involves imaging mechanical properties of tissues by incorporating aspects of acoustics and wave propagation into medical imaging systems. Elastography techniques have been developed across many existing imaging modalities. The first report that introduced the concept of noninvasively mapping tissue physical responses to mechanical excitatory forces with magnetic resonance imaging (MRI) was by Muthupillai et al. in 1995, who termed the method magnetic resonance elastography [1]. Thereafter, Manduca et al. 2001 described many of the principles underlying the mechanics, imaging technique, and imaging processing of MR elastography, which has since served as the basis for two decades of advances aimed at improving the accuracy, reliability, and clinical application of the technique in many different tissues [2]. Since it was first developed, MR elastography has been applied to many human tissues including liver, kidneys, spleen, pancreas, breast, heart, cartilage, prostate, lungs, brain, and muscles [3–5]. Clinically, the first utilization of MR elastography was in patients with hepatic fibrosis [6].

Utilization of MR elastography for investigating human brain in both healthy and pathological states has gained more attention over the past several years. Elastography has revealed systemic changes in the viscoelastic properties of the brain during aging and with several neurodegenerative disorders. These mechanical changes are reflective of underlying degradation in tissue microstructure [7, 8]. This growing interest has led to further refinement and continued evolution of brain MR elastography techniques, with modern approaches en vogue demonstrating high reliability and reproducibility with remarkably low test–retest variability of mechanical property estimates [9]. There is some data now available on global and regional brain tissue shear properties in healthy adult brains [10]. Deep gray matter structures are noted to be stiffer than surrounding white matter tracts, and global cerebrum stiffness is higher than that of the cerebellum [11–14]. There is preliminary data to suggest that brain stiffness also appears to decrease regionally both with normal aging and dementia [15, 16]. Additionally, regional mechanical property measurements are associated with measures of brain function in young adults [17]. Translation of those novel brain structural–functional relationships to potential meaningful therapeutic interventions is currently being pursued [18].

While brain MR elastography methods and applications in the adult brain have matured, with several upcoming clinical applications, there is a conspicuous lack of brain MR elastography data in children. There are only a few published reports of pediatric brain mechanical properties and potential applications in pediatric neuropathology [19–22]. The intent of this article is to provide a guide for clinicians and clinical researchers to adopt brain MR elastography in their pediatric neuroimaging protocols. We review brain MR elastography technique, describe the options currently available for clinical use, and discuss the existing and potential applications where brain MR elastography may have an impact.

Brain MR elastography technique

The MR elastography technique involves three main components: a mechanical actuator that induces the required shear wave deformation throughout the tissue, a pulse sequence that encodes the resulting motion into the MRI phase images, and an inversion algorithm that calculates the shear modulus distribution from the imaged wave field (Fig. 1). From the complex shear modulus, comprising the storage and loss moduli, it is common to report the magnitude modulus or the shear stiffness, both of which capture how hard it is to deform the tissue, and are more dependent on tissue composition and elastic properties [23]. The viscous properties are often reported as the damping ratio or phase angle, which are dimensionless ratios of viscosity to elasticity, and are commonly considered reflective of tissue organization [24].

Fig. 1.

Overview of brain MR elastography technique. MR elastography imaging sequences encode brain tissue motion generated by vibrating the head, and the resulting wave fields are used to estimate tissue mechanical properties. Stiffer tissues exhibit longer wavelengths while softer tissues exhibit shorter wavelengths. MR elastography property maps include the complex shear modulus and the shear stiffness, which is related to wave speed in a viscoelastic material

Although brain MR elastography uses the same principles as in other body organs, such as the liver, the nature and complexity of the tissue architecture and mechano-protective role of the skull require special considerations to reliably study the mechanical features of the brain. MR elastography data needs to be collected with sequences that can capture wave displacements in three dimensions and across the entire brain, with high resolution and signal-to-noise ratio. The tissue mechanical properties are then determined from the imaged waves using an inversion algorithm that can capture the highly heterogeneous nature of brain tissue.

Mechanical actuation and actuator hardware

Waves in the brain are generated by external actuators vibrating the patient’s head, typically applied at a single frequency. The objective in choosing a vibration frequency is to have short wavelengths (i.e., at higher frequencies) but with sufficient displacement amplitude at the center of brain after attenuation (i.e., at lower frequencies). Balancing these two factors has led to 50 Hz and 60 Hz as the most common vibration frequencies for brain MR elastography in both adults and children [7–9, 19–21]. Some centers use multiple applied frequencies in the same patient to improve precision and quality of acquired data [25, 26]. The motion applied to the skull is typically small, on the order of 20–50 microns, while the amount of tissue deformation is an order of magnitude smaller [27]. The amount of vibration in brain MR elastography was compared with a safety standard for vibration exposure and it was found that the amount of applied vibration in elastography is very small relative to the amount of vibration exposure considered safe, thus brain MR elastography has been considered a safe technique in that light [28]. A prospective study evaluating the safety of brain MR elastography in 20 healthy adult volunteers (mean age 36.5 years, range 25–53 years) concluded that the small level of motion encountered in short brain MR elastography experiments performed based on standards for human exposure to vibrational energy is safe [29]. Studies rigorously examining safety of MR elastography on children have not yet been performed. However, experience at our center in children as young as five years old indicates that amount of motion transmitted to the brain is similar as in adults (i.e. similar amount of encoded wave motion in the brain), thus we consider the safety studies described above to apply to children as well as adults.

External actuator systems are composed of an active driver for generating motion and a passive driver for transmitting the shear waves into the target tissue. As this technology has advanced and matured, pneumatic actuators have become the most common for the brain and other organs as they allow for ease of patient setup and enhanced flexibility in reconfiguration of the actuator for different clinical applications. A widely-used, FDA-approved pneumatic actuator is manufactured by Resoundant, Inc. (Rochester, MN) and can be purchased through all major MRI scanner vendors. The active driver is the same as what is used for liver MR elastography but the passive driver used for brain MR elastography is a soft “pillow” actuator at the posterior of the patient’s head [7–9]. This driver is easy to set up and can be in place during the entire imaging session (Fig. 2).

Fig. 2.

Soft pillow driver commonly used for brain MR elastography with the Resoundant pneumatic actuator system in a standard head coil

Imaging strategy

MR elastography relies on acquiring vibration-induced tissue displacement and deformation data by encoding to the phase of the MRI signal using motion encoding gradients [30]. The gradients are applied sequentially in three orthogonal directions of the image plane and synchronization with vibration is shifted in time to capture temporal wave propagation. A key difference with MR elastography of the liver, which often has used gradient-echo-based imaging, is that the most common pulse sequence for brain elastography has been spin-echo, echoplanar imaging (SE-EPI) [9, 31, 32]. This technology has allowed for robust, whole-brain MR elastography images with a 2–2.5 mm isotropic voxel size and sufficient signal-to-noise ratio.

Efforts to develop imaging sequences capable of achieving even higher resolution have often used multi-shot spiral imaging, although these sequences have some practical challenges for clinical translation given the need for a separate image reconstruction that can take significant time [34–37]. Alternatively, several approaches have been proposed to reduce the total number of images necessary to be acquired to make a complete MR elastography dataset by manipulating how motion encoding gradients are applied, thus shortening acquisition time [38–40]. Active research to improve brain MR elastography involves sampling and reconstruction to further shorten scan times and methods to correct for motion artifacts, both of which may be particularly important for pediatric applications [35, 41–44].

EPI-based MR elastography sequences are available on all main scanner platforms. Advanced MR elastography sequences providing new capabilities for brain applications are available from scanner vendors as work in progress (WIP) sequences or from researchers developing these sequences. FDA approval specifically for brain MR elastography is in process.

Mechanical property estimation

The final aspect of an MR elastography experiment is the calculation of brain mechanical properties from imaged displacement data using inversion algorithms [45]. While there are several different types of inversion algorithms used in MR elastography for different applications, direct inversion algorithms are the most common, involving algebraic solution of the simplified wave equation without significant computational requirements [2]. Direct inversion is a robust and useful tool in assessing tissue stiffness, however it has some drawbacks for use in the brain, including extra sensitivity to noise, reduced resolution due to the homogeneity assumption, and outliers in estimated properties. Direct inversion algorithms are currently widely available and are included as part of online image processing that comes standard with brain MR elastography sequences.

Other MR elastography inversion algorithms are a consistent topic of research as developers seek to improve the accuracy, precision, and resolution of recovered property maps. Nonlinear inversion is a popular approach for brain MR elastography and involves using a finite element implementation of the tissue mechanics as a forward model to iteratively solve for the complex shear modulus by optimizing differences between measured and calculated displacement fields [46–49]. However, the nonlinear inversion algorithm can be computationally expensive and typically requires sufficient computing resources, making it more technically challenging and time consuming to use. The neural network inversion is a promising algorithm that uses machine learning methods to estimate mechanical properties and this framework offers opportunities to improve property estimation by training on data that better matches brain mechanical behavior [50–52]. Future developments may make neural network inversion a powerful tool for brain MR elastography as it becomes more widely available. In the cases of nonlinear inversion, neural network inversion, and other advances or recently developed inversion algorithms, they are generally available primarily for research purposes and not yet part of standard acquisition and reconstruction packages. Nevertheless, clinical researchers are encouraged to adopt such approaches depending on their specific applications.

Comparison with ultrasound elastography

Ultrasound elastography may also prove to be useful in examining brain tissue mechanical properties in pediatric populations [53, 54]. Ultrasound elastography may be potentially advantageous in very young populations, including neonates, where initial studies have already been published, but MR elastography studies have not yet been performed [55]. Ultrasound elastography of the brain is not a mature field, and comparisons with MR elastography are lacking, but may yield complementary value for specific applications.

Normative brain MR elastography data in the pediatric brain

As previously stated, to date there are only several reports of brain mechanical properties in pediatric populations [19–21]. In a cohort that included 36 healthy children by Yeung et al., average whole brain white matter stiffness was not significantly different between children, adolescents, and adults (2.15 vs. 2.24 vs. 2.33 kPa, respectively) [20]. Similarly, in a study of 26 children and adolescents aged 7–17 years, Ozkaya et al. found no significant increases in stiffness with age, however they found a significant increase in white matter viscosity of about 6.9% across that age range [21]. In another study by McIlvain et al., comparing brain stiffness in adolescents with young adults, it was also shown that average whole brain stiffness was similar between the two groups (3.13 vs. 3.23 kPa for adolescents vs. young adults), but individual structures including the cerebellum, the caudate, and the putamen were significantly different [19]. A later study showed that brain mechanical properties differ across development, with brain stiffness decreasing from age five to adulthood, where the magnitude of change with maturation varied significantly between brain regions [56]. It was additionally shown that the difference in stiffness between the nucleus accumbens and prefrontal cortex, representing reward and control systems, was related to risk taking behavior in adolescents [57]. These studies point to regional changes in pediatric brain stiffness occurring as the brain matures into adulthood, and that these properties have behavioral associations making them useful in understanding the developing brain. Longitudinal studies and robust comparisons of MR elastography with other imaging modalities, such as diffusion tensor imaging, have not yet been conducted but are necessary to more completely understand how tissue mechanical properties relate to structural maturation and cognitive function. Additionally, studies that examine brain mechanical properties in children less than five years of age, effects of puberty, and associated sexual dimorphism in brain mechanics are critical to understanding normal tissue properties in this pediatric populations.

Clinical applications of MR elastography in the pediatric brain

Pediatric hydrocephalus

In adults, MR elastography has been used to study patients with normal pressure hydrocephalus, which is typically associated with progressive cognitive decline. An early study found brain tissue in normal pressure hydrocephalus patients had significantly lower stiffness than normal controls (2.27 vs. 2.84 kPa; −25.1% lower in patients), with periventricular regions exhibiting the lowest stiffness, which is presumably due to tissue damage in these regions as this stiffness did not recover after surgical shunting [58, 59]. A more recent report by another group found seemingly contradictory patterns of increased temporal, parietal, and occipital lobe stiffness in a cohort of patients with normal pressure hydrocephalus compared to age and gender matched controls (2.5% to 12.0% higher in patients) [60]. Follow up research with advanced inversion algorithm methods suggested that there is likely an agreement between the earlier studies, with patients showing both periventricular softening and peripheral stiffening patterns [52].

Regarding pediatric hydrocephalus, a study on 39 patients who received a shunt at a very young age (less than 2 years) for early onset hydrocephalus, Wagshul et al. performed MR elastography scans years later (9–36 years old) and found that patients that had received a shunt for hydrocephalus exhibited 6.9% lower stiffness throughout the white matter compared to age-matched healthy controls without hydrocephalus [61]. Importantly, this study found that lower stiffness in patients was also associated with poorer quality of life and higher likelihood of depression [61]. Of note, low pressure hydrocephalus in children is a rare condition of variable etiology that can develop when cerebrospinal fluid drains as surgical shunts are set to a negative pressure [62]. An interesting case of a 19-year-old female presenting with low pressure hydrocephalus was reported where the patient exhibited very low brain stiffness – approximately 46% softer than a normal healthy brain [63]. Upon follow-up two years later her brain stiffness was noted to be back to near normal [64]. This case further points to the complex biomechanics of hydrocephalus that could be studied with MR elastography to perhaps aid in the management of these patients [65].

Pediatric idiopathic intracranial hypertension

In animal models of idiopathic intracranial hypertension, a neurological condition characterized by raised intracranial pressure, stiffness measurements quantitatively appear to correlate with the magnitude of acute intracranial pressure changes [66]. Recent brain MR elastography reports showed viscoelasticity changes in brain tissue induced by physiological Valsalva maneuver in healthy adult volunteers, supporting the potential use of MR elastography for examining patients with idiopathic intracranial hypertension [67]. In a limited sample of young adults with idiopathic intracranial hypertension, MR elastography revealed slightly higher brain stiffness in patients compared to controls, but with no immediate significant stiffness changes after treating those patients with lumber puncture procedures, suggesting that elevated intracranial pressure was not solely responsible for the differences in brain stiffness [68]. A recent study by Cogswell et al. in a larger sample of adults (n = 30) agreed with these findings: higher stiffness in idiopathic intracranial hypertension but without significant improvement either acutely or after six months [69]. Studies using ultrasound-based, shear wave elastography in neonates and children with increased intracranial pressure have demonstrated potential correlation with intracranial pressure values, however, reports of MR elastography-based studies in a similar cohort of patients are not yet available [70, 71]. Nevertheless, it is plausible to consider expanding the currently recruiting clinical trials to include children to better understand MR elastography utility of studying pediatric idiopathic intracranial hypertension [72].

Pediatric cerebral palsy

Cerebral palsy (CP) is a neurodevelopmental disorder of motor function and muscle tone, which is attributed to an early life injury to the developing brain [73, 74]. Chaze, et al., studied CP using brain MR elastography and found a significant reduction of mechanical integrity of both gray and white matter in children with CP compared with age-matched controls, despite all study participants having only minimally affected motor function and no associated structural brain lesion [31]. In a follow-up on the same cohort, McIlvain et al., reported that greater brain stiffness in many primary locomotion regions is indicative of greater performance on an assessment of dynamic balance abilities [75]. These studies suggest MR elastography is sensitive to neuropathology in CP that affects clinical outcomes of interest. Brain MR elastography may be implemented in routine evaluation of children with CP to better understand the progression of disability and response to therapeutic interventions (Fig. 3).

Fig. 3.

Brain MR elastography findings in children with cerebral palsy which exhibited lower brain stiffness compared to age-matched typically-developing children. Data originally presented by Chaze et al. [31]

Pediatric brain tumors

Noninvasive evaluation of pediatric brain tumor mechanical properties has recently gained attention, and studies have indicated stiffness-dependent behavior of brain tumors in terms of development, invasion, spread and treatment response [76–78]. In adults, MR elastography has been utilized in the context of either presurgical evaluation of tumor stiffness and consistency to guide resection (such as in brain meningiomas), or in the realm of tumor grading and prognostication (such as in brain gliomas) [26, 79–81]. In preliminary adult MR elastography data, shear stiffness was lower in high grade gliomas than in unaffected brain tissue [26]. MR elastography of brain tumors of glial origin demonstrated direct correlation between the degree of softening and tumor grade, as well as with certain genetic alterations, though a similar pattern in pediatric brain tumors, which can be fundamentally different than tumors in the adult brain, is yet to be studied and established [82–84]. Preliminary data from our center on low grade, pediatric gliomas, suggests utility of brain stiffness in revealing tumor characteristics (Figs. 4, 5, 6 and 7) [85].

Fig. 4.

Axial anatomical image and MR elastography stiffness map from a 13-year-old female with low grade glioma (marked by arrow)

Fig. 5.

Axial anatomical image and MR elastography stiffness map from a 11-year-old male with low grade glioma (marked by arrow)

Fig. 6.

Axial anatomical image and MR elastography stiffness map from a 16-year-old female with low grade glioma (marked by arrow)

Fig. 7.

Axial anatomical image and MR elastography stiffness map from a 9-year-old female with low grade glioma (marked by arrow)

Pediatric multiple sclerosis

Pediatric multiple sclerosis (MS) is a chronic demyelinating disease with a typical onset in children or young adults which has a well-defined, child-specific diagnostic criteria that heavily rely on neuroimaging characteristics [86]. Though there is limited information on MR elastography in pediatric MS, in a study of 45 adult MS patients, overall cerebral stiffness was lower by an average of 13% compared to matched healthy volunteers [87]. These biomechanical alterations appear to also differ with the clinical and radiological subtypes of MS. For instance, adults with chronic progressive MS showed a pronounced reduction in cerebral shear stiffness compared to relapsing–remitting variant and suggested the potential benefit of monitoring disease progression and anticipation of relapses in adult MS [88]. Animal models of MS have revealed the likely microstructural basis of these stiffness alterations and have demonstrated a consistent correlation between changes in stiffness and macrophage and microglial infiltration as assessed by F4/80 gene expression studies, as well as softening due to induced demyelination [89–91].

Regional alterations in brain stiffness have also been observed in other related brain demyelinating disorders. In neuromyelitis optica spectrum disorders, accelerated softening and reduction of stiffness was noted in the thalamus and related white matter tracts in adults [92]. In clinically isolated syndrome, an often pre-MS diagnostic entity occasionally seen in children, patients exhibited significant reduction in brain stiffness by more than 15% in most regions [25]. Although diagnostic criteria differ for pediatric MS [93], it is plausible that similar patterns of brain mechanical properties maybe seen in children with MS and other demyelinating disorders.

Pediatric epilepsy

Focal cortical dysplasia, a highly epileptogenic lesion and a frequent cause of drug resistant focal epilepsy in children, is a neuronal migration disorder which is thought to have different anatomical and physical continuants compared to surrounding healthy tissues [94]. An expanding utilization of shear wave elastography in patient undergoing focal cortical dysplasia surgery is emerging, although these are primarily adults-based trials [95]. Chan et al., reported successful localization of a 0.5 cc lesion that was stiffer than surrounding tissue prior to resection using intraoperative ultrasound-based, shear wave elastography consistent with small focal cortical dysplasia in a 7-year-old child, which was not detected on prior 3 T MRI, B-mode ultrasound, ictal SPECT, or PET studies [96]. Additionally, in a sample of adult patients with mesial temporal lobe epilepsy, which is the most common form of refractory epilepsy, hippocampal stiffness measured via brain MR elastography was found to be significantly higher compared to healthy adult volunteers and MR elastography was able to differentiate patient from control groups effectively [97]. Of note, a recent pediatric clinical trial studied stiffness using novel intraoperative tonometer device in 24 children who underwent epilepsy surgery and concluded that increased brain stiffness was significantly associated with the presence of MRI lesion, severity of cortical disorganization, and recent subdural grid implantation, with real-time measurements to aid in identifying brain lesions [98].

Conclusion

Magnetic resonance elastography is a novel, rapidly evolving, non-invasive, well-tolerated, quantitative magnetic resonance imaging technique for assessing the viscoelastic properties of the brain. Although there are only limited published studies on application of brain MR elastography in pediatric population, there is a growing body of evidence suggesting the potential role of MR elastography in various pediatric neurological disorders such as hydrocephalus, demyelinating disease, and epilepsy. There are additional potential applications where MR elastography studies are currently lacking, such as ischemic or traumatic brain injury which likely have effects on brain tissue integrity and mechanical properties.

Continued advances in faster data acquisition strategies and other advances in methods tailored to pediatric populations offer a promise of seamless incorporation of MR elastography in clinical workflows, which will further enable studies to establish and solidify the use of MR elastography in clinical pediatric populations. We also note however there is a complete lack of studies in children younger than five years old and as such much less is known about the mechanics of the brain in this age. It is also critical to rigorously consider safety as we expect motion transmission may be greater in children under five years of age.

Data Availability

There is no data associated with this manuscript.