Physiological and pathological modulations of intrinsic brain activity assessed via resting-state fMRI

Since its inception in 1992, functional magnetic resonance imaging (fMRI) has considerably boosted our knowledge of the human brain function, primarily due to its non-invasive nature, and its relative high spatial and temporal resolution. Among the available fMRI contrasts, blood-oxygenation level-dependent (BOLD) signal plays a leading role in this field. The contrast is based on the different magnetic properties of the haemoglobin which - combined with the specific relation existing between neuronal, vascular and metabolic activity - allows to ascribe variations in the measured signal to variations in the underlying neuronal activity. During BOLD acquisitions, the comparison of different cognitive states in task-based experiment (alternating rest states to sensory or cognitive stimulations) has revealed the modular organization of the human brain function, an operation that is commonly referred to as functional brain mapping. Surprisingly, task-induced activity requires an increase in brain’s energy consumption by less than 5 percent of the underlying baseline activity. Most of the brain’s energy demand, from 60 to 80 percent, is used to sustain intrinsic, task-unrelated, neural activity (Raichle, 2006). In this light, functional brain mapping, utilizing task-based fMRI, focuses only on the tip of the iceberg, whereas most of the brain’s activity remains largely uncharted. The notion that the brain has an intrinsic or spontaneous activity is known from early electro-encephalography (EEG) measures due to Hans Berger. However, only in recent years, after the seminal work of Biswal and colleagues (Biswal et al., 1995), the study of spontaneous brain activity has overwhelmingly emerged as a primary field of research in neuroscience. In the so called resting-state condition (i.e., when the brain is not focused on the external world), Biswal reported BOLD low-frequency (< 0.1 Hz) fluctuations (LFFs) synchronized across functionally related and anatomically connected regions. Thereafter, several studies have consistently shown that specific patterns of synchronized spontaneous LFFs identify different resting-state networks, including, but not limited to, visual, motor, auditory, and attentive network. The overall picture emerging from thousands of resting-state fMRI studies depicts a never-resting brain, continuously engaged in maintaining communications within several wide-distributed networks. Such intrinsic brain activity, reflected in spontaneous BOLD LFFs, is the focus of the present thesis. The study of LFFs in spontaneous BOLD signal can reveal much about brain’s functional organization, especially considering that signal variability has been related to variability in behaviour (Fox et al., 2007). In addition, the simplicity of data acquisition – subjects just lie in the scanner refraining from falling asleep - makes the technique particularly suited for studying pathological conditions, in which subject’s cooperation might not fulfil the demands of task-based studies. Indeed, several psychiatric and neurological disorders, including degenerative dementia, have shown altered patterns of LFFs, even in the absence of observable anatomical abnormalities (Barkhof et al., 2014). Thus, how the intrinsic brain’s activity is modulated in response to different behavioural states and in response to pathological conditions can give insights into the brain functionality and into the mechanisms behind illnesses, respectively. Importantly, correct result interpretation is highly influenced by the type of metrics adopted and how they are implemented. The resting-state approach to the study of the brain’s function has required the development of more sophisticated processing and analysis techniques compared to those commonly applied in task-based fMRI. While seeking for task-responding regions in the brain is guided by information embedded in the experimental paradigm, in steady-state fMRI no a priori cue is provided. In such experiment the extraction of relevant information is based on (i) the temporal synchronization between spatially segregated elements of the brain, feature known as functional connectivity, and on (ii) the amplitude of the oscillation per se, a measure of the strength of the intrinsic brain activity. Despite such simple classification, the field of resting-state fMRI is scattered with a disparate amount of metrics, each of which highlight different facets of spontaneous LFFs. Before turning to the study of spontaneous LFF modulations, we will provide a comprehensive and optimized mathematical framework for the extraction of relevant information from resting-state data (Chapter 2). The results of this effort is an easy-to-use matlab toolbox specifically designed for the processing and analysis of steady-state fMRI data. In principle, the information coded in functional connectivity and in oscillation amplitude are unrelated. While the former assesses the degree of cooperation between segregated elements of the brain, the latter quantifies the neural workload of each single brain’s element, independently from the activity of other regions. Nonetheless, modulations in both measurements have been reported in several pathological conditions - yet in separate studies - suggesting a possible relation between them. In this context, we sought to investigate the potential coupling between the functional connectivity and the oscillation amplitude in cohort of healthy elderly and the probable modulations induced by dementia of the Alzheimer’s type (Chapter 3). Regardless of how the brain relates the two types of measures extractable from resting-state data, their disease-induced modulations are relevant per se in uncovering the illness. Indeed, Alzheimer’s disease is known to produce alterations in spontaneous brain activity, both at the synchronization and the amplitude level (Wang et al., 2007). Since the hallmark of the pathology is a profound deficit in episodic memory, much effort has been done in characterizing the alterations in spontaneous brain activity underlying such deficit. Contrarily, little is known about another commonly reported deficit, the language related impairment (Taler and Phillips, 2008). In the second part of Chapter 3 we sought to disclose the brain regions underpinning language deficits by looking at the alterations in functional connectivity of the relevant network. While the study of LFFs in pathological conditions can contribute to reveal the mechanisms behind the pathology and how it spreads into the brain, the study of spontaneous brain activity in physiological conditions can disclose the intrinsic brain functionality. In healthy subjects the resting brain has been extensively characterized and its network topology has shown to be a consistent and reliable physiological feature (Damoiseaux et al., 2006). An intriguing issue is how the brain reorganizes its patterns of spontaneous BOLD LFF while it is focusing on the external world. Indeed, the intrinsic brain activity is not an exclusive feature of the resting condition, instead it is present also on the top of the task-evoked response. In chapter 4, with peculiar experimental paradigms we separated the task-evoked response from the intrinsic brain activity during sustained cognitive stimulations. In a first experiment we sought to characterize the spatio-temporal proprieties and the dynamic of the transition from a resting to a stimulated condition. In the second part we specifically investigated how the brain reorganizes its internal functional architecture during visuospatial attention. Indeed, besides strongly affecting the processing of visual incoming stimuli, visual spatial attention also affects brain networks. Recent studies suggest that visual attention affects functional connectivity within and between the visual network and the attention network (Spadone et al., 2015), yet modulations of attention on brain networks are still poorly understood.