Surfing Early Electromagnetic Waves: Unveiling the Developing Brain with EEG/MEG

Biomagnetometers based on SQUID technology are currently still the only available technology to assess simultaneously fetal brain and heart development in utero with a high temporal resolution. This high temporal resolution is essential for deciphering the neural processes underlying higher order processing in fetal life. This is crucial to increase our understanding of developmental trajectories especially related to the concept of 'Developmental Origin of Health and Disease'. During the development of healthy fetuses we were able to demonstrate that they can show long-term memory-associated responses during a local-global paradigm during the last trimester. Data of this study also provided evidence that neural complexity unexpectedly decreases during the last trimester also showing sex specific differences. A possible explanation of this finding is that the analysis was related to event related responses and not spontaneous activity. The understanding of detours of developmental trajectories in fetuses affected by different stressors including psychological, metabolic and/or environmental stressors are necessary to also provide possible intervention possibilities. Metabolic stressors, e.g maternal Gestational Diabetes or psychological stress affect the developmental trajectory of fetal brain and heart development. First evidence also indicate that these stressors have to be taken into account in the preconceptional state. To further advance research in this direction new cost-efficient recording possibilities, e.g. optical pumped magnetometers are crucial and we will discuss the current status in the field. 

During the third trimester of gestational age, with the consolidation of thalamo-cortical connectivity, the refinement of neural circuits shifts from being guided primarily by endogenous neural activity to being guided also by exogenously induced neural activity. Evidence suggests that external sensory information makes a major contribution to the neuroplastic changes underlying neurodevelopment. We will present a unique cohort of 80 premature newborns, born between 27 and 37 weeks of gestational age, recorded with high-density electroencephalography (EEG) during rest and auditory rhythmic stimulation. We will present: (i) the systematic evolution of phase-amplitude coupling and aperiodic oscillatory characteristics of spontaneous neural activity; (ii) the development of neural synchronization to auditory periodicities (an important characteristic of musical rhythm and speech prosody); and (iii) the maturation of neural capacities for integrating auditory information across distributed cortical networks. In addition, we will show that integrating the characteristics of the neural response induced by exogenous stimulation and those of spontaneous endogenous neural activity can give reliable prediction of cerebral age across the third trimester of gestation. Finally, we will present and discuss a recording setup to measure newborn brain activity with Optically Pumped Magnetometers (OPM-MEG) and present first ever recordings from preterm infants during rest and auditory stimulation. Our studies provide new insights into how the assessment of the developmental dynamics of neuronal networks at rest and simultaneously of their distributed response to exogenous stimuli could constitute a key element in evaluating neurodevelopment and adaptation of the premature newborn to its environment. 

In the present talk, I will focus on multisensory integration (MSI) from a developmental perspective. In the first study I will present you, we aimed at investigating whether MSI is already present and spatially organized at birth. By recording EEG responses to unimodal (audio and tactile) and bimodal (audio-tactile) stimulation in 25 human newborns, we described a genuine electrophysiological pattern of MSI at birth, with super-additive responses to bimodal stimuli. Furthermore, MSI was spatially modulated by the proximity to the baby’s body, with larger MSI effect when auditory stimuli occurred close to the body district receiving tactile stimuli. This suggests that the prenatal experience of multisensory cues during the long and sensory-enriched gestation characterizing human pregnancy likely allows MSI to emerge even before birth. Thus, in the second study I will present you, we moved our research question backward to the prenatal life and we asked when MSI and its spatial organization emerge along the third trimester of gestation. We capitalized on foetal eye-lens movements recording through 2D ultrasound, as a reliable measure of foetal attention orienting we successfully described in response to visual, acoustic and vibro-tactile stimuli delivered in isolation. In the MSI protocol, we deliver unimodal (audio and visual) and bimodal (audio-visual) stimulation at different timepoints (i.e. 27, 33, 37 weeks). Visual stimuli are delivered at a fixed location according to the foetal head position, while acoustic stimuli are delivered either colocalized or not colocalized with the visual ones. Once MSI is present and spatially organized, we should expect to observe an increased number of foetal eye-lens movements in response to bimodal than unimodal stimuli and to colocalized than not colocalized bimodal stimuli. 

Electroencephalography (EEG) is a standard tool for brain monitoring in the neonatal intensive care unit, although traditionally only with limited channel montages and all clinical interpretation performed manually by eye. In research environments, there has been significant progress in developing high-density electrode montages and machine-learning-based algorithms for automated interpretation. These enable mapping of brain activity across the cortex, extraction of information not easily discerned visually, and the prospect for new diagnostic and prognostic biomarkers. This talk will present our recent advances in computational analysis of neonatal EEG. This includes estimating functional brain age as a marker of maturation, automated seizure detection, measuring large-scale functional network activity in infants at risk of adverse neurodevelopment, and assessing the extent to which personalized MRI-derived cortical surfaces affect EEG source reconstruction versus standard template brains. Challenges in translating these tools to the clinic will also be discussed.