An accumulating amount of evidence supports the hypothesis that various brain disorders are in fact disconnection syndromes or network diseases, which disrupt brain networks that support multiple perceptual and cognitive functions. These include mild cognitive impairment, dementia, depression and schizophrenia, as well as epilepsy, and also several other neurological and psychiatric diseases. Although the biological foundations of many of these disorders remain largely unknown, there is now general consensus that they can manifest themselves in observable brain changes, and that these are not restricted to specific brain areas but rather distributed across the whole brain. The most promising approaches in the quest for related biomarkers are based on using neuroimaging, mostly fMRI, but also EEG, to map whole-brain networks during rest (namely the so-called resting-state networks, RSN’s) and extract measures of the underlying intrinsic functional connectivity. However, one major concern is that the haemodynamically-driven changes underlying the BOLD signal measured with fMRI are caused not only by neuronal but also non-neuronal mechanisms, rendering the BOLD-fMRI technique vulnerable to a number of confounds.
One important approach providing further insights into the neuronal component of fMRI measurements is its combination with concurrent EEG recordings, which more directly reflect neuronal activity (Jorge et al., NeuroImage 2014; Murta et al., Hum. Brain Mapp. 2015). The multimodal integration of EEG and fMRI also takes advantage of the highly complementary spatiotemporal properties of the two types of data, and simultaneous EEG-fMRI has been actively sought over the past two decades, particularly for studying spontaneous brain activities such as those occurring during resting states or in relation with epilepsy. However, a number of challenges must still be overcome in order to make the most out of simultaneous EEG-fMRI. On the one hand, improved techniques are required to deal with the severe artefacts induced on the EEG in the fMRI environment, and a better understanding of the EEG and fMRI neurophysiological correlates is needed to develop better methods for integrating the two signals. On the other hand, it is crucial to unequivocally identify the synchronous activity arising from spurious sources of no interest in the BOLD signal, the so-called physiological noise, in order to obtain BOLD-fMRI measurements of purely neuronal origin. Significantly, the growing use of ultrahigh fields (notably, 7 Tesla) in search of improved sensitivity may be compromised in resting-state EEG-fMRI studies. Not only do EEG artefacts aggravate considerably with field strength, potentially hindering the benefits of its multimodal integration with fMRI, but also physiological noise amplitude increases with the MRI signal and thus field strength, imposing a limit on the temporal SNR that can be effectively achieved.
We have been addressing both of these challenges, including the development of improved methods for the minimization and correction of EEG artefacts in the fMRI environment, with special impact at 7 Tesla (Jorge et al., NeuroImage 2015; Rothlübbers et al., Brain Topogr. 2014), the reduction of dimensionality and extraction of relevant information from the EEG data (Abreu et al., J. Neurosci. Methods 2016; Marques et al., Hum. Brain Mapp. 2009), and the investigation of biophysically-informed EEG metrics to predicted BOLD changes (Leite et al., Front. Neurol. 2013; Murta, Hu et al., NeuroImage 2016; Murta, Chaudhary et al., NeuroImage 2016). We have also been interested in modelling and removing physiological noise in order to preserve the expected sensitivity benefit of 3D, vs. 2D, fMRI acquisition schemes at 7 Tesla (Jorge et al., Magn. Reson. Imag. 2013; Abreu et al., NeuroImage 2016). Moreover, we have addressed the problem of using EEG-fMRI to map brain networks and model their functional connectivity, with the purpose of inferring about causality in the propagation of epileptic activity (Murta et al., NeuroImage 2012). We have also been studying spontaneous fluctuations in functional connectivity, occurring on short time scales of seconds or minutes, and how they may be related with the dynamics of physiological and brain dynamics, as well as pathology.
Principal Investigator: Patrícia Figueiredo