The functional connectivity between brain regions of intrinsic brain networks such as the default-mode network (DMN) is known to be increased. Patients with several neuropsychiatric diseases show alterations in resting-state networks, e.g. in Alzheimer’s disease (Franciotti et al. 2013). Studies using simultaneous EEG-fMRI measurements investigated resting-state activity with respect to BOLD correlates of EEG oscillations of specific frequencies such as alpha-oscillations (Goldman et al. 2002; Laufs et al. 2003; Moosmann et al. 2003). Beyond that, the EEG signal has also been related to the functional connectivity within and between networks in order to characterise network dynamics. For example, delta and beta oscillations explain 70 % of the DMN variance of functional connectivity (Hlinka et al. 2010). An increase of alpha power leads to a decrease of fMRI connectivity between the primary visual cortex and occipital regions as well as a decrease of the negative coupling between visual areas and regions of the DMN (Scheeringa et al. 2012). The strength of connectivity between DMN and the dorsal attention network (DAN) has been found to be inversely correlated with alpha power, whereas alpha power correlated with the spatial extent of anticorrelation between DMN and DAN (Chang et al. 2013).

FMRI patterns have also been related with complex patterns of EEG organisation such as so-called microstates. These are quasi-stable and unique topographic representations of the EEG reflecting the summation of concurrent neuronal activity across brain regions rather than being specific for a certain frequency band (Lehmann et al. 1987). Recently, the relationship between EEG microstates and BOLD resting-state networks has been investigated using simultaneous EEG-fMRI (Britz et al. 2010; Huang et al. 1996; Musso et al. 2010). Microstates are known to be altered in different psychiatric disorders such as schizophrenia (Lehmann et al. 2005).

The relationship between EEG coherence and fMRI connectivity patterns has been investigated by Wang et al. using the combination of fMRI and invasive electrophysiological recordings. This study revealed that the coherence in low frequencies (<20 Hz) predominantly contributes to fMRI connectivity patterns in the resting state in monkeys (Wang et al. 2012). However, especially high-frequency oscillations (30–100 Hz) seem to be tightly linked to the BOLD signal (Mukamel et al. 2005; Niessing et al. 2005). In order to integrate findings from EEG coherence and fMRI connectivity in humans in the future, the combination of simultaneous EEG-fMRI and EEG source analysis represents a promising approach.

Scalp EEG activity includes oscillations of a variety of distinct frequency ranges each of which is supposed to represent the synchronisation of varying neuronal networks. The rhythmic activities in the resting or “spontaneous” EEG are usually divided into several frequency bands (delta, <4 Hz; theta, 4–8 Hz; alpha, 8–12 Hz; beta, 12–30 Hz; and gamma, >30 Hz) which are associated with different behavioural states, ranging from sleep to relaxation, heightened alertness and mental concentration (Lindsley 1952; Niedermeyer and Lopes Da Silva 2004; Nunez 1995). A comprehensive analysis of the different types of brain rhythms can be found in Buzsáki’s monography Rhythms of the Brain (Buzsaki 2006).

Several studies investigating the relationship between fMRI and special oscillatory EEG components reported especially high correlations between high- in comparison with low-frequency electric activity and BOLD signal. Recordings in cats revealed that correlations between LFPs and BOLD signal are especially high in the gamma-band frequency range (Niessing et al. 2005). Similar results were found in studies combining fMRI and intracranial EEG recordings in human epilepsy patients (Mukamel et al. 2005) showing also a predictive value of fMRI about the anatomical location of cross-condition gamma-band modulations (Brovelli et al. 2005; Lachaux et al. 2007). Foucher et al. reported a differential reactivity of event-related potentials (ERPs) and BOLD signal in response to the presentation of two kinds of rare events (i.e. target and novel stimulus in an oddball task) using combined measurement of EEG and fMRI. However, in accordance with fMRI results, target-related gamma oscillations were more intense than their novel-related counterparts (Foucher et al. 2003). The authors of this study propounded a physiological hypothesis underlying these findings: as opposed to ERPs reflecting synchronous activity of synapses within milliseconds (Baillet et al. 2001; Speckmann and Elger 1999) and similar to the BOLD signal, event-related gamma oscillations are less dependent on the small jitter of the neuronal response relative to a stimulus because they do not require to be time-locked to the stimulus that precisely (Tallon-Baudry and Bertrand 1999). Furthermore, ERPs reflect the synaptic input function of pyramidal cells only (Baillet et al. 2001; Speckmann and Elger 1999), whereas fMRI reflects the synaptic activity of all neural cells including inhibitory interneurons (Logothetis et al. 2001; Mathiesen et al. 2000; Matsuura and Kanno 2001) which are involved in the synchronisation of gamma oscillations (Traub et al. 1996), too. A third possible reason could be related to the assumption that ERPs might correspond to the simple phase resetting of ongoing cerebral activity (Makeig et al. 2002) which should not consume much energy (Foucher et al. 2003).

Using the EEG-informed fMRI approach, Mobascher et al. investigated the influence of nicotine on the EEG P300 component of the visual evoked potential elicited by an oddball-type task in patients with schizophrenia. The authors report a nicotine-associated increase in P300-informed fMRI activation in schizophrenia patients and healthy controls, mainly in the anterior cingulate cortex (ACC) and adjacent medial frontal cortex. The results of separate analyses of EEG and fMRI measures were reported to be largely unaffected by nicotine (see Fig. 4.2) (Mobascher et al. 2012). Musso et al. conducted an EEG-fMRI study on the human ketamine model of schizophrenia and report diminished P300 amplitudes (visual oddball task) and corresponding BOLD responses in the ketamine condition in cortical regions being involved in sensory processing and selective attention (Musso et al. 2011). The EEG-informed fMRI approach has recently also been used to describe the brain network involved in reward processing (Hauser et al. 2013; Plichta et al. 2013).

With respect to ongoing brain rhythms in the resting state, Balsters et al. identified neuroanatomical origins of EEG measures with high sensitivity to the age-related cognitive decline and their manipulation by the acetylcholinesterase inhibitor donepezil. Investigating oscillatory patterns with an EEG-informed fMRI approach, the authors found drug-induced changes of delta, alpha and beta band power associated with activity differences within the hippocampus (delta, see Fig. 4.3), frontal-parietal network (alpha) and default-mode network (beta) (Balsters et al. 2011). In a study investigating resting-state networks with simultaneously recorded EEG-fMRI patients with a schizophrenia spectrum disorder showed lower EEG frequencies to be related to the default-mode and left-working-memory network compared to healthy controls (Razavi et al. 2013).

Based on recent advances in EEG-fMRI, a first successful approach of real-time integration of simultaneous fMRI and EEG has been made. Thus, it was possible to perform a simultaneous multimodal real-time fMRI and EEG neurofeedback study: participants were able to simultaneously regulate their BOLD fMRI activation in the left amygdale and frontal EEG power asymmetry in the high beta band. These results indicate possible applications of the EEG-fMRI approach in enhanced cognitive therapeutic strategies for major neuropsychiatric disorders (Zotev et al. 2014).