Methods for functional connectivity and morphometry in neonatal neuroimaging to study neurodevelopment.

Preterm birth is a major pediatric health problem that perturbs the genetically determined program of corticogenesis of the developing brain. As a consequence, prematurity has been strongly associated with adverse long-term neurodevelopmental outcome that may persist even into adulthood. Early characterization of the underlying neuronal mechanisms and early identification of infants at risk is of paramount importance since it allows better development of early therapeutic interventions aiming to prevent adverse outcomes through resilience. This dissertation aims to investigate the consequences of preterm birth on brain function and structure and their relation to adverse neurodevelopmental outcome, as well as to unveil the effect of an early music intervention on brain function. Research to date has mainly focused on the effect of early interventions on the long-term outcome but not on the effect of those interventions on brain function in preterm populations. Moreover, a clear consensus about the predictive utility of MRI volumetric data for long-term outcomes is still missing.

This dissertation consists of three main parts. First, we investigate the effects of an early musical intervention in preterm infants using a task-based and resting-state fMRI paradigm. We explore brain effective connectivity changes during task and functional connectivity (FC) modulations at rest. We show that music exposure during hospitalization promotes brain auditory processing maturation. We find increased FC between the auditory cortex and regions involved in familiarity and music processing, such as the thalamus and caudate nucleus. Moreover, we report increased rs-FC in regions involved in associative memory and multisensory processing, such as the calcarine and angular gyrus with a dosage-dependent effect on this modulation. Secondly, we examine brain volumetric development and whether brain volumes are associated with cognitive and behavioural long-term outcomes. We find that perinatal risk factors, such as asphyxia and sepsis have an impact on brain tissue volume. We also find that cognitive and behavioural outcomes are strongly associated with parental SES and that brain tissue volumes at TEA have moderate predictive power. Next, inspired by studies that have characterized the progression of neurodegenerative diseases, we employ an advanced technique to unveil the sequence of biological events along the course of prematurity. Our novel findings might be of clinical use as they allow more targeted early interventions aiming to improve long-term outcomes. Finally, we propose a framework that enables the spectral graph theory analysis of the functional connectome at the voxel level. Our results indicate the presence of both functional gradients and community structure revealing fine-grained connectivity patterns. The proposed framework opens a new avenue for spectral graph theory research at the voxel level.

Real-Time fMRI Neurofeedback: Methods and Application to Chronic Tinnitus

Nicolas Gninenko

In the last twenty years, advances in real-time functional magnetic resonance imaging (rt-fMRI) have offered exciting new tools to study the human brain. One of them, termed rt-fMRI neurofeedback (NF), has turned the MRI scanner environment into an interactive brain-computer interface (BCI), allowing to position brain activity as the independent variable in the relationship ranging from brain function to observed behavior. This paradigm immediately gained considerable interest in the fMRI research community, as the idea of grasping volitional control over a precisely targeted brain area suddenly became feasible. Since then, a multitude of studies have used this technique in drastic efforts to normalize pathological brain activity (or connectivity) in a wide range of neurological disorders, ultimately seeking clinical and behavioral improvement.Among these, chronic tinnitus, commonly referred to as "ringing in the ears", has been considered a promising condition for NF interventions. Tinnitus has been linked to wide disruptions in brain functions involving attention, cognition, emotional distress, and auditory and sensory processing, among others. It has also been associated with hyperactivity in the primary auditory cortex after acoustic trauma. Currently, no cure exists for a complete relief of symptoms. A previous proof-of-concept study suggested that learned down-regulation of the auditory cortex slightly improved short-term subjective reports of tinnitus severity in a small group of participants. Here, we present the longest NF clinical trial to date in chronic tinnitus (NeuroTin), in which 60 patients are enrolled across 3 different interventions: electroencephalography NF (EEG NF), fMRI NF, and cognitive behavioral therapy (CBT), nowadays considered as standard care for tinnitus. The aim is to assess whether any of the two NF approaches can outperform standard CBT training for alleviation of tinnitus distress. In this dissertation, we present clinical outcomes and neuroimaging findings for the fMRI NF group.Twenty-one participants learned to down-regulate the activity of their primary auditory cortex over 15 weeks of fMRI NF training. While only a few participants managed to learn down-regulation, we found a significant clinical relief in tinnitus distress lasting up to one year in the fMRI NF group, as opposed to the CBT group. We demonstrate the superiority of fMRI NF training over standard care CBT and show, for the first time, resulting long-term benefits in quality of life. Furthermore, we bring additional evidence regarding the implication of a highly modulated small brain region in tinnitus, bilaterally located in the parietal operculum (OP3). We describe its effective connectivity with regions pertaining to the auditory pathway and discuss its implication in a possibly altered multisensory integration mechanism in tinnitus. Then, using a co-activation patterns analysis, we unveil 4 functional brain networks co-occurring with the deactivation of OP3. We identify them as (1) an auditory-visual network comprising an important node of the auditory pathway (the medial geniculate body), (2) a part of the NF control network, (3) the default mode network, and (4) a somatosensory network likely linked to other sensory integration modalities. Lastly, we characterize the average dynamics of these 4 functional networks and discuss new avenues for fMRI NF interventions and treatments in chronic tinnitus.

EPFL2022

Spatio-temporal coupling of the electric and hemodynamic brain responses in humans

Roberto Martuzzi

Our comprehension of human brain functions and their dynamics has been dramatically improved by recent developments in non-invasive imaging techniques. These methods can be divided into two different categories, according to the nature of the measured signal: hemodynamic techniques, such as functional magnetic resonance imaging (fMRI) and positiron emission tomography (PET), and electromagnetic techniques, such as electroencephalography (EEG) and magnetoencephalography (MEG). These two categories are have complementary characteristics: hemodynamic techniques have a good spatial resolution (on a millimeter spatial scale) but have a poor temporal resolution, which is inherently limited by the rate changes in blood flow and oxygenation. Electromagnetic techniques have sub-millisecond temporal resolution but have a poor spatial resolution, since the analysis of intracranial generators requires the solution of an underdetermined inverse problem (i.e. there are infinite solutions that can explain equally well the same scalp-recorded distribution). The complementarity of the characteristics of these two families of methods allowed researchers to suppose that the understanding of spatio-temporal brain dynamics can be drastically improved by their combination (so-called multimodal imaging). Unfortunately some caveats hinder such combination. First, the nature of neurovascular coupling is still poorly understood. Second, analytical methods for multimodal imaging are largely in their infancy. The first part of this thesis focuses on the analysis of the temporal characteristics of the blood oxygenation level dependent (BOLD) signal and on how they are modulated by stimulus conditions. To analyze the BOLD dynamics, a novel method for synchronizing stimulus delivery and volume acquisition was developed. This method allows for estimating the BOLD signal with a high temporal resolution (in this thesis up to 125 ms) and for studying how the temporal characteristics (in this thesis mainly the BOLD peak latency and slope) are modulated by stimulus conditions (with an approach similar to that used in the analysis of the EEG evoked potentials). We applied this novel technique to a simple reaction time task to lateralized visual stimuli (the so-called Poffenberger paradigm) as well as to a multisensory auditory-visual reaction time task. In the first study (the Poffenberger paradigm) the analysis of BOLD dynamics supported the theory of a bilateral visuo-motor pathway even in the case of a visual stimulus ipsilateral to the responding hand. In the second study, (the auditory-visual multisensory reaction-time task), the analysis showed auditory-visual interactions within both primary auditory and visual cortices that could not be otherwise revealed by traditional fMRI analysis methods since it does not involve changes in signal amplitude. The second part of this thesis focuses on the comparison of the statistical results obtained by the analyses of fMRI and of the intracranial local field potentials (LFPs), estimated by the ELECTRA inverse solution. We first developed a new method for the analysis of EEG data. This method is based on the statistical comparison of the spectral characteristics of the estimated intracranial LFPs of the pre- and post- stimulus onset periods. Each single trial is analyzed independently, without including an averaging step, so that the information carried by high frequencies is preserved. We also propose a new metric, called resemblance, to investigate the relationship between fMRI and the estimated intracranial LFPs. Single-trial analysis and the resemblance metric were applied in an experiment involving separate EEG and fMRI acquisitions during the same passive visual stimulation protocol. This experiment revealed that only a limited set of LFP frequencies shows a spatial correlation with fMRI. This set of frequencies changes across brain areas, such that progression from lower to higher cortical levels of visual processing incorporates at each step new frequencies. In conclusion, in this thesis we show that the estimation and the analysis of the BOLD time course can give an important contribution to better understanding brain functions and brain organization. To fully understand the meaning of changes in BOLD dynamics, we need a better knowledge of the neuro-vascular coupling. To do that, we introduced a new method for evaluating the relationship between EEG and fMRI across frequencies and anatomical regions.

EPFL2006

Principal components of functional connectivity: A new approach to study dynamic brain connectivity during rest

Nora Leonardi, Jonas Richiardi, Dimitri Nestor Alice Van De Ville

Functional connectivity (FC) as measured by correlation between NM BOLD time courses of distinct brain regions has revealed meaningful organization of spontaneous fluctuations in the resting brain. However, an increasing amount of evidence points to non-stationarity of FC; i.e., FC dynamically changes over time reflecting additional and rich information about brain organization, but representing new challenges for analysis and interpretation. Here, we propose a data-driven approach based on principal component analysis (PCA) to reveal hidden patterns of coherent FC dynamics across multiple subjects. We demonstrate the feasibility and relevance of this new approach by examining the differences in dynamic FC between 13 healthy control subjects and 15 minimally disabled relapse-remitting multiple sclerosis patients. We estimated whole-brain dynamic FC of regionally-averaged BOLD activity using sliding time windows. We then used PCA to identify FC patterns, termed "eigenconnectivities", that reflect meaningful patterns in FC fluctuations. We then assessed the contributions of these patterns to the dynamic FC at any given time point and identified a network of connections centered on the default-mode network with altered contribution in patients. Our results complement traditional stationary analyses, and reveal novel insights into brain connectivity dynamics and their modulation in a neurodegenerative disease. (C) 2013 Elsevier Inc. All rights reserved.