Pediatric Magnetoencephalography in Clinical Practice and Research

INTRODUCTION

Magnetoencephalography (MEG) is a noninvasive neuroimaging technique that measures the electromagnetic fields generated by the human brain. The main sources of these fields are considered to be the postsynaptic currents in the apical dendrites of the cortical pyramidal cells,¹ although action potentials also may be recorded.^2,3 MEG offers an excellent temporal resolution in the range of submilliseconds, that allows even brain activity with high frequencies to be recorded.⁴ MEG also offers good spatial resolution via magnetic source imaging (MSI).^5,6 It is most sensitive to the currents that are tangential to the surface of the scalp and blind to magnetic fields produced by radial sources.^7,8

In clinical practice, MEG is used mostly for the localization of epileptogenic foci in patients with drug-resistant epilepsy undergoing surgery.^9,10 MEG also has become popular for understanding the human brain in both adults and children.¹¹ During an MEG recording, the patient or subject sits comfortably in an armchair that is located inside a magnetically shielded room, which is specially designed to prevent the electromagnetic noise of the environment from overwhelming the weak neural electromagnetic fields. In several occasions, simultaneous high-density electroencephalogram (EEG) recordings are performed (Fig. 1). Recording also can be performed in a supine position when a patient is under sedation or needs to sleep. The magnetically shielded room is equipped with adjustable lighting and an audiovisual communication system that allows communication with the technical staff seated outside. Brain activity is measured by positioning the patient head inside a special helmet that contains magnetic field detection coils, which are inductively coupled to very sensitive magnetic field detection devices, called superconducting quantum interference devices (SQUIDs). In order to operate, SQUID devices must be in an extremely low temperature, close to absolute zero. For this reason, SQUIDs are placed inside a thermo-shielded tank that is filled with liquid helium. Modern MEG systems accommodate a high number of coils up (up to 306).

Fig. 1.

Simultaneous MEG and high-density EEG recordings from a 4 year-old-girl. Recordings were performed at Cook Children’s Health Care System, Fort Worth, Texas.

For decades, MEG was used mostly for measuring the brain activity of adults, and its use in pediatric clinical practice and research was limited. This may be explained by the fact that only adult MEG systems were available in the market. These systems have a fixed-size helmet that is relatively large for pediatric heads, particularly for infants and young children. Yet, several researchers used these systems for pediatric studies (for a review of these studies, see Chen and colleagues¹²) by placing the children’s heads either in the center of an adult MEG helmet or in such a way that the brain area of interest was as close to the sensors as possible. But this placement results in large distances between the active neural generators and the MEG sensors in infants and young children, who have significantly smaller heads compared with the adults. Such a distance leads to less than optimal and sometimes even inadequate signal-to-noise ratio for infant MEG recordings.¹³

This article highlights the benefits that pediatric MEG has to offer to clinical practice and pediatric research, particularly for infants and young children; reviews the existing literature on using adult MEG systems for pediatric use; briefly describes the few pediatric MEG systems currently extant; and draws attention to future directions of research, with focus on the clinical use of MEG for patients with drug-resistant epilepsy.

PEDIATRIC MAGNETOENCEPHALOGRAPHY SYSTEMS

To address the limitations discussed previously, MEG systems specially designed for pediatric use have been developed. These systems are equipped with a helmet having smaller dimensions in order to reduce the distance between the neural generators and MEG sensors. It has been claimed that such a design provides greater sensitivity and spatial resolution compared with adult systems,¹⁴ although these have not yet been quantified. Examples of these systems are (1) the SQUID Array for Reproductive Assessment (SARA) (VSM Med-Tech, Port Coquitlam, British Columbia, Canada) (Fig. 2A, B)¹⁵ that is being used for fetal and infant MEG recordings; (2) the babySQUID system (Tristan Technologies, San Diego, CA) that has a partial head coverage^10,14 (Fig. 2C); (3) the KIT whole-head system (model PQ1151R) (Yokogawa/KIT, Tokyo, Japan) (Fig. 2D); (4) the Artemis 123 whole-head system (Tristan Technologies, San Diego, CA) (Fig. 2E); and the (5) MagView whole-head system (Tristan Technologies, San Diego, CA)¹⁶ (Fig. 2F). A detailed description of these systems is provided elsewhere.¹²

Fig. 2.

MEG systems specially designed for pediatric use. (Adapted from Chen Y-H, Saby J, Kuschner E, et al. Magnetoencephalography and the infant brain. Neuroimage 2019: 189: 445–458.)

CLINICAL APPLICATIONS OF PEDIATRIC MAGNETOENCEPHALOGRAPHY

The main clinical application of MEG is in the presurgical evaluation of patients with drug-resistant epilepsy. These patients undergo extensive screening using numerous neuroimaging techniques before their surgery. The goal of this screening is the localization of the epileptogenic zone, the brain area that is indispensable for the generation of seizures. MEG is used predominantly for the localization of the irritative zone, the brain area that generates interictal epileptiform discharges (IEDs), and for the localization of the eloquent cortex, the cortical areas that are responsible for the different normal functions. It has been shown in 1000 patients with refractory epilepsy that complete resection of the irritative zone localized with MEG is associated with significantly higher chances of achieving seizure freedom in the short term and the long term.¹⁷ These findings show that MEG provides nonredundant information, which significantly contributes to patient selection, focus localization, and ultimately long-term seizure freedom after epilepsy surgery.

The localization of the irritative zone involves the initial identification of the IEDs in the MEG traces, the modeling of the electromagnetic properties of the head and of the sensor array (i.e., the forward model), and finally the estimation of the brain sources, which produced the interictal spikes according to the head model in question (i.e., the inverse problem). MEG source localization also requires coregistration between a patient’s anatomy and MEG sensors’ locations. MEG spikes are chosen for analysis based on duration (<80 ms), morphology, field map, and lack of associated artifact (Fig. 3). It has been shown that MEG can localize the irritative zone with an accuracy of approximately 15 mm.¹⁸ The forward problem estimates the magnetic field measured outside the scalp from a known distribution of neural activity generators using Maxwell’s electromagnetic equations. Since the introduction of powerful computers, realistic head models have been used, such as the boundary element model or finite element model, for estimating the forward problem. The inverse problem estimates the most probable neural activity in a child’s brain that can explain the MEG signals, which are recorded outside the human’s head. The inverse problem does not have a unique solution and requires a priori knowledge about the source generators to constrain the inverse problem solution. Such a constraint assumes that all MEG activity recorded outside the head can be modeled by an infinitesimally small line-current element, the equivalent current dipole (ECD). The ECD is the only validated and approved method for clinical use in the presurgical planning of epilepsy patients.^19–21 It is assumed that each interictal discharge is generated by one ECD. The task is to find the location and direction of this ECD (either at the peak or at the onset of each spike) and overlaid onto the patient’s own coregistered brain magnetic resonance image (MRI). This process of modeling the IEDs and overlaying them on patient’s MRI often is referred to as magnetic source imaging, MSI. The desirable scenario is a tight cluster of dipoles with similar orientation in a focal brain area (≥20 spikes in a 1-cm area); each of these dipoles represents one interictal discharge (Fig. 4). It has been reported that approximately 90% of patients in whom the MEG cluster is completely resected achieve seizure freedom 1 year after resective surgery, whereas only 25% of patients with a partial resection of the MEG cluster attain seizure freedom.²² In addition, patients are significantly more likely to achieve seizure freedom when the stereotaxic EEG sampling completely covered the MEG clusters. On the other hand, MEG spike sources may have relatively poor correlation with the seizure-onset zone (SOZ), the brain area where seizures are initiated, that is regarded as the best estimator of the epileptogenic zone.²³

Fig. 3.

IEDs in a 4-year-old girl with history of prolonged febrile seizures and febrile focal-onset seizures with gradual impairment of awareness and ultimately motor involvement, including some clonic activity of either side. Display of magnetometers covering the left frontal (top) and left temporal (bottom) areas. Topography indicates a change of magnetic flux in the left frontotemporal areas. Recordings are performed using the MagView^™ system that accommodates 375 magnetometers in a helmet that is specially designed for children up to 4 years old. No sedation was administered during the recording. The recording was performed while the patient was awake.

Fig. 4.

MSI findings of IEDs identified in Fig. 3. Localization is performed through ECD. Dipoles with goodness-of-fit greater than 60% were considered and displayed. Dipole cluster is localized in the vicinity of Brodmann area 45 (BA45). Right and left indicates left and right hemispheres. Both images depict a 3D representation of the patient’s brain (the left is on three slices: axial-coronal-sagittal; the right is on a cortical surface).

Fig. 5 shows MEG of a pediatric patient with tuberous sclerosis and intractable epilepsy. His MRI showed numerous cortical and subcortical tubers throughout his cerebral hemispheres and numerous subependymal nodules; any of the tubers could have been an epileptic focus. The MEG showed that all his interictal spikes originated from 1 tuber in the right posterolateral temporal lobe. This provided valuable information to the neurosurgeons, guiding them on which tuber to resect, to decrease frequency of seizures.²⁴

Fig. 5.

Pediatric patient with tuberous sclerosis and multiple bilateral cortical and subcortical tubers throughout his cerebrum and with intractable epilepsy. The MEG found 83 interictal spikes over 50 minutes of recording, which all localized to a tuber in the right posterolateral temporal lobe. This aided the neurosurgeons in determining which tuber to resect. (Courtesy of Dr R Lee, MD, San Diego, California.)

Several studies have shown that MEG also can be used for the localization of the SOZ, the brain area where the seizures are initiated.^25–27 The SOZ is regarded as a more precise estimator of the epileptogenic zone than the irritative zone. Ictal MEG in children demonstrates good concordance with the SOZ, as defined by the current gold standards: intracranial EEG and surgical outcome.²⁸ Yet, it is challenging to capture a seizure during an MEG scan due to the fact that seizures do not occur all the time and MEG recording time is limited (typically less than an hour) in the clinical setup. Moreover, ictal MEG recordings have low signal-to-noise ratio because ictal activity is obscured from the movement artifact that accompanies the seizure.²⁹

Finally, pediatric MEG has been used for the localization of the eloquent cortex in children with refractory epilepsy undergoing surgery. MEG is gaining increasing acceptance for the noninvasive mapping of language^30,31 as well as motor and sensory areas^32–34 compared with other neuroimaging techniques. This mostly is because methods based on electromagnetic measures of brain activity have the advantages of being insensitive to the distortive effects of anatomic lesions on brain microvasculature or metabolism on the developing brain³⁵ and providing a less intimidating recording environment for younger children.³¹ Many of the MEG language studies use similar paradigms, consisting of spoken nouns and a list of target nouns, which must be identified.³⁶ Little published work has been undertaken on optimizing language protocols specifically for the pediatric population. One of the greatest challenges for MEG with young individuals is to ensure there is minimal movement. To achieve this, scanning procedures need to remain short whilst ensuring adequate signal to noise.

PEDIATRIC MAGNETOENCEPHALOGRAPHY FOR UNDERSTANDING HUMAN BRAIN DEVELOPMENT

Although MEG has been used widely for assessing the brain activity in adults over the past 3 decades, it still is rarely used in developmental research. This has been in part due to the fact that dedicated infant and young children whole-head MEG systems have been developed only recently. In addition to excellent temporal and spatial resolution, discussed previously, the advantages of MEG for measuring neural activity in infants and young children include (1) minimal distortions of volume current caused by incompletely developed fontanels and sutures,³⁷ resulting in more accurate source modeling than EEG; (2) head movement compensation methods that correct for use with awake infants during cognitive and sensory tasks; and (3) more accurate source analyses that are possible with age-appropriate MRI templates, mitigating the need for individual MRIs. This section provides an overview of the role of dedicated pediatric MEG systems in developmental cognitive neuroscience research. Given that the critical period of brain development happens early in life, in this section, the review focuses on existing infant MEG studies on basic sensory processes to high-level cognitive processes from birth to 4 years of age. Specifically, studies of the following research fields are discussed: (1) primary auditory processes to simple stimuli; (2) somatosensory processes; (3) visual processes; (4) speech and language processes; and (5) spontaneous resting state brain rhythms.

Auditory Processes in Infants and Young Children

Researchers have utilized MEG to track development of primary sensory auditory responses before and after birth.^15,38–43 Using fetal MEG systems (i.e., SARA), researchers examining auditory evoked responses in fetuses and neonates reported a steady decrease of P2m latency as a function of gestational age. Such studies are unique to MEG, because MEG signals are less distorted by amniotic fluid and layers of skin and muscle. After birth, studies have shown that the latency of the auditory evoked field (AEF) continues to decrease for several years after birth,^44,45 with a slower rate over the first few months of life. Using dedicated whole-head pediatric MEG systems, researchers were able to examine activity in source space and separately examine the development of hemispheric lateralization of AEF. In older children, studies have shown earlier right than left auditory response latency at 50 milliseconds (ms).⁴⁶ The hemisphere differences in the maturation of left and right auditory cortex in infants and young children still remain unclear. Edgar and colleagues⁴⁴ showed that across children ages 6 months to 59 months, auditory P2m latency decreases at a rate of approximately 0.6 ms/mo, with right hemisphere auditory encoding advantages observed only under more demanding encoding conditions. Longitudinal studies following children from infancy through toddler-hood to school age are needed to better understand the hemisphere maturation rate of infant and young child AEF and how it is associated with later left hemisphere lateralization for word processing that emerges during reading acquisition.⁴⁴ Looking to the future, multimodal imaging studies in pediatric populations will provide direct examination of associations between brain neural function and brain structure to better understand the mechanisms that support maturation of different brain regions and the development of language. As a first step, however, given the limitation and lower success rate of obtaining structural or functional MRI in pediatric populations, MEG is the most optimal neuroimaging modality to measure whole-brain neural networks associated with basic and high-level cognitive processes in children.

Somatosensory Processes in Infants and Young Children

Research examining primary/secondary somatosensory cortical activity in children typically employ painless pneumatic tactile stimuli.^33,34 Somatosensory evoked fields (SEFs) in newborns traditionally were measured in response to tactile stimulation applied to the fingertip of sleeping infants positioned against one side of a conventional adult-sized MEG helmet.^47–51 Infant SEFs were observed as one broad and slower deflection approximately 60 ms poststimulus, localized in the contralateral primary somatosensory cortex (SI). In adult literature, SEFs were characterized as two deflections from SI. The morphology differences of SEFs between infants and adults might be due to possible GABAergic inhibitory processes, which are not yet maturing in infants.⁵² Regarding clinical application of infant SEFs, several studies have suggested that SEFs localized in secondary somatosensory cortex (SII), peaking at approximately 200 ms poststimulus, may be a prognostic brain marker for predicting outcome in preterm infants^53,54 or infants born with prenatal drug exposure.⁵⁵ Other studies have focused on developmental changes of sensorimotor brain rhythms, such as mu-rhythm peak frequency,^52,53 as well as functional connectivity measures of sensorimotor neural network.

Visual Processes in Infants and Young Children

Few studies have used MEG to study primary visual evoked fields (VEFs) or evoked responses or brain responses to social or face stimuli in young children. Using a train of light flashes, studies have shown decreased VEFs to successive light flashes as a sign of habituation of visual responses in fetuses and newborns.^43,56 Although neuroimaging studies examining visual processes or socio-cognitive processes using visual stimuli (e.g., face) have been done in preschool-aged or school-aged children,^57,58 MEG studies examining how infants process social visual stimuli have not yet been conducted.

Speech and Language Processes in Infants

Most of the MEG studies examining how language is processed during the first year of life have focused on localizing language-related brain regions beyond primary auditory cortex.^59–63 Compared with other neuroimaging techniques, MEG is unique in that it can be used to study higher-level cognitive processes (e.g., language processes) throughout the whole brain in awake infants and young children. Using distributed source modeling (for example, minimum norm estimation⁸), Broca area and cerebellum have been found to be involved in passive encoding speech sounds as well as speech production in addition to auditory cortex.^60,61 A growing literature of bilingual processes in infants also showed that prefrontal and orbitofrontal areas are involved during passive listening of speech sounds in bilingual but not monolingual infants, suggesting that dual language exposure during infancy may be related to the development of executive function skills later in life.^64–66

Resting-State Brain Rhythms in Infants and Young Children

Characterizing spontaneous brain rhythms in awake infants and young children is challenging. Characteristics of low-frequency brain rhythms also have been studied as potential indicators of brain pathology in adult literature. As such, studies characterizing slow rhythms early in life can be used as a potential brain marker for infants at risk for developmental disorder. For example, Sanjuan and colleagues⁶⁷ showed associations between theta power and maternal stress in infants, suggesting a delay in cortical maturation in infants born to mothers with posttraumatic stress disorder.

FUTURE DIRECTIONS

Efficacy and safety of pediatric epilepsy surgery have been significantly improved over the past decades.⁶⁸ Only a third of pediatric surgical candidates, however, proceed to surgery within 2 years of onset, despite this onset having occurred at less than 2 years of age in 60% of the children.^69,70 Recent progress in pediatric MEG has provided a comprehensive surgical management for these patients,^10,71 although its use is limited. Application of MEG in young children with epilepsy will accelerate during the coming years, as different types of pediatric whole-head MEG systems and more advanced data analysis methods become available to researchers and clinicians. These advances will lead to greater use of MEG as a complement to clinical EEG, with improved noninvasive delineation of the epileptogenic zone. Research examining brain neural activity in pediatric populations using EEG is sizable, and studies examining patterns of brain blood flow in infants using functional MRI are increasingly prominent. As this review indicates, MEG is a promising noninvasive technology for studying infant and young children’s brains, offering complementary information for understanding the development of brain function beyond brain structure in pediatric populations, especially in infants and young children. Recent advances in MEG analyses, particularly in the use of age-matched MRI templates instead of individual MRIs, will facilitate analyses of brain function in source space, thereby providing greater potential to study brain networks and functional connectivity in pediatric MEG research. With emerging hardware as well as analysis pipelines dedicated specifically to pediatric populations that provides necessary sensitivity, future pediatric MEG research and clinical studies likely will apply distributed source localization to examine activity throughout the brain, and thus facilitate studies of local and long-range functional connectivity. Longitudinal as well as cross-sectional studies are needed to evaluate the developmental trajectory (maturation) of neural activity in the first few years of life. It is expected that departures from neurotypical trajectories will offer early detection and prognosis insights in infants and toddlers at risk for neurodevelopmental disorders, thus paving the way for early targeted interventions.

SUMMARY

During the past 10 years, MEG has become increasingly useful for the presurgical delineation of epileptogenic zones and eloquent cortex in both lesional and nonlesional pediatric cases. Several studies also have used pediatric MEG to study brain development, particularly at early years of life during critical periods. Application of MEG in pediatric epilepsy and research of human development will accelerate during the coming years as different types of pediatric whole-head MEG systems and more advanced data analysis methods become available to researchers and clinicians.

KEY POINTS.

• Magnetoencephalography (MEG) is a noninvasive neuroimaging technique that measures the electromagnetic fields generated by the human brain.

• During an MEG recording, the patient or subject sits comfortably in an armchair that is located inside a magnetically shielded room, which is specially designed to prevent the electromagnetic noise of the environment from overwhelming the weak neural electromagnetic fields.

• The magnetically shielded room is equipped with adjustable lighting and an audiovisual communication system that allows communication with the technical staff seated outside.

• Brain activity is measured by positioning the patient head inside a special helmet that contains magnetic field detection coils, which are inductively coupled to very sensitive magnetic field detection devices, called superconducting quantum interference devices.