Neurodynamic measures of functional connectivity and cognition

Abstract

Exploring the structure and function of the human brainrelies on a range of non-invasive measures and computa-tional approaches. Neurodynamic functional brain imagingmethods offer unique insight into the human brain pro-cesses by capturing neuronal activity in real time. Mea-surements of scalp potential differences of spontaneous andevoked activity were the first electrophysiological measurethat revealed differences in the measured waveforms whichwere state, stimulus, task, context etc. dependant. Fourdecades ago the application of ultrasensitive supercon-ducting quantum interference device (SQUID) for themeasurements of extremely week magnetic fields of thehuman brain led to a dynamic development of magnetoe-cephalography (MEG) research [2]. Development of thisnew functional brain imaging method resulted in openingnew questions and furthering the analysis of the EEG/ERPas well since the interest of MEG researchers moved fromthe analysis based on the evaluations of the measuredwaveforms to explore the approaches to identify and trackcortical activations in space and time. Spatio-temporalsource localization models and inverse bio-electro-mag-netic problem estimation approaches were developed toestimate the underlying neuronal substrates of the evokedactivity related to sensory perception, movement, andcognition. Exploration of the underlying dynamic corticalnetworks of the working human brain was made possiblenot only in the measurement space but also at the func-tional source level.During the 1990s, The Decade of the Brain, non-inva-sive structural and functional brain imaging methods wereregarded as revolutionary in exploring the structure andfunction of the human brain in healthy and diseased states.The wealth of data regarding sensory, motor, and cognitiveprocesses provided initially by positron emission tomog-raphy (PET) is dominated during the last two decades by anexplosion of research using functional magnetic resonanceimaging (fMRI). fMRI is often reported to offer superbmillimeter spatial resolution, almost uniform throughoutthe brain. EEG and MEG were, however, regarded asmodel dependent and consequently prone to erroneoussource location estimation. Multi-modal integrationappeared in mid 1990’ as a solution to overcome ill-posednature of the bio-electro-magnetic problem and a low, 2 sand even 40 s, temporal resolutions of fMRI and PET,respectively. The initial confidence that PET or fMRIidentified functional foci could be used to constrain EEG orMEG solution was soon reconsidered.Unlike in EEG and MEG, there is a great degree ofstandardization regarding preprocessing and statisticalapproaches in fMRI. During a rather intensive data gath-ering period since its discovery important questions such asthe reliability of the fMRI results remain almost uncon-sidered and unanswered [3]. Another open question is therelationship of the hemodynamic and/or metabolicresponse and the neuronal activity of interest. Despite awidely used terminology referring to PET and fMRI data asneuroimaging their data represent only correlates of theneuronal activity. Consequently, there is an increasingawareness that all noninvasive functional brain imagingmethods are in fact subject to solving inverse problems tomove from the measurement space to the source space in