We review here studies mapping BOLD changes related with multiple EEG phenomena, conspicuous during cognitive, sensory, or motor functions: ERP morphological characteristics, EEG synchronisation and phase coherence, and EEG power fluctuations within multiple frequency ranges.
Mapping BOLD changes related to ERP amplitude and latency
Several studies using scalp EEG and fMRI data simultaneously acquired have investigated single-trial correlations between the amplitude and/or latency of ERP and the amplitude of the BOLD signal (Benar et al., 2007; Debener et al., 2005; Eichele et al., 2005; Fuglo et al., 2012; Mulert et al., 2008). For example, Mulert et al. (2008) investigated the spatial distribution of statistically significant correlations between fluctuations in the amplitude of the N1 potential (AofN1), measured during a forced choice reaction task, under low- and high- effort conditions, and the amplitude of the BOLD signal. By contrasting AofN1-related BOLD changes under high- effort and passive listening, the group found positive correlations in the anterior cingulate cortex. Mulert et al. (2008) showed that single-trial correlations could be particularly helpful to separate different aspects of the BOLD signal based on their specific correlation to different ERP features (e.g.: N1 potential fluctuations related to the high-effort condition).
Mapping BOLD changes related to EEG synchronisation and phase coherence
Single-trial scalp EEG synchronisation and phase coherence have also been investigated as potential predictors of the BOLD signal amplitude (Kottlow et al., 2012; Mizuhara et al., 2005). For example, the first study analysing the BOLD correlates of common-phase signals during visual binding in humans, Kottlow et al. (2012), searched for (40 - 42 Hz) global field synchronisation (GFS) - related BOLD changes, during film viewing. The film consisted of face parts changing their positions, which, during some periods, rearranged themselves, yielding to a visually recognisable face (FACE). A unitary boxcar regressor (FACE) and a scaled boxcar regressor (modulated by the GFS values) (GFS-FACE) were convolved with a double- HRF. The GFS-FACE regressor was orthogonalised with respect to the FACE regressor, in order to be able to test for the presence of GFS-dependent modulations of the BOLD signal. The group found positive GFS-BOLD correlations in the bilateral middle fusiform gyrus and left precuneus, important regions for visual binding and face perception.
Mapping BOLD changes related to the EEG spectral content in the δ, θ, α, β, γ ranges
The BOLD correlates of the power within the classical EEG frequency bands have also been investigated during tasks (Michels et al., 2010; Mulert et al., 2010; Scheeringa et al., 2011a, 2011b). For example, Michels et al. (2010) investigated the relationship between the amplitude
of the BOLD signal and the power of θ (5 - 7 Hz), (low-: 8 - 10 Hz; high-:10 - 12 Hz), β (low- β: 13 - 20 Hz; high-β 20 - 30 Hz), and γ (30 - 40 Hz) bands, during the retention phase of a working memory task in humans. All frequency bands were included in a single GLM. The group found only positive correlations for high-β (20 - 30 Hz) in dorsolateral prefrontal cortex (DLPFC), and inferior frontal gyrus (IFG); and for γ in IFG, and medial prefrontal cortex (MPFC). The group also found only negative correlations for θ in MPFC, posterior parietal cortex (PPC), cingulate cortex (CC); and for high- (10 - 12 Hz) in parieto-occipital regions. For low- (8 - 10 Hz), and low-β (13 - 20 Hz), both positive and negative correlations were found, in diverse locations. They concluded that the power of both low and high frequency bands correlates with the amplitude of the BOLD signal, in diverse locations; and that these correlations tend to be negative for the lower frequencies (<7 Hz) and positive for the higher ones (>20 Hz). While investigating γ (40 Hz) amplitude – related BOLD changes during an auditory task, Mulert et al. (2010) found positive correlations in the auditory cortex, thalamus, and anterior cingulate cortex. Furthermore, Scheeringa et al. (2011a), who investigated the relationship between the power of multiple frequency ranges and the amplitude of occipital BOLD changes in healthy subjects performing a visual attention task, found positive single-trial correlations for the high-γ (60 - 80 Hz) range, and negative single-trial correlations for (10 Hz) and β (18 - 28 Hz) ranges. Interestingly, Scheeringa et al. (2011b) compared the amplitude of the BOLD signal for visual stimuli given at the peak or trough of the cycle, and found a stronger positive BOLD response for stimuli given at the peak, which suggests that the phase of the rhythm at which the stimulus is given has an impact on the simultaneous BOLD response.
To this point, we reviewed studies mapping the BOLD correlates of multiple EEG-derived features, over the entire brain. Notwithstanding the relevance of these studies, one might be interested in the more general questions “Which aspects of the electrophysiological signal best correlate with the amplitude of the BOLD signal?”, “Are these aspects common across regions and brain states?”. Aiming to address the first question, at least, in part, Rosa et al. (2010) compared several heuristic EEG-derived metrics in the form of moments of the EEG spectrum (in the range 1 - 40 Hz) in terms of their individual capability to explain occipital BOLD fluctuations recorded in healthy subjects performing a visual task; they found that the amplitude of task- related BOLD changes was best explained by the root mean squared frequency function, proposed by Kilner et al. (2005), 𝑞 𝑅𝑀𝑆𝐹(𝑡) =√∑ffmaxmin𝑓2𝑃(𝑓, 𝑡), which revealed more significant voxels and higher statistical significance levels.
In summary, the electro-haemodynamic coupling function during both non-epileptic spontaneous activity (§ 2.2.1.2) and cognitive, sensory, and motor functions is likely to be a complex combination of the power and frequency of the electrophysiological signal. The studies reviewed to this point are somewhat unsatisfactory due to the intrinsic limitations of scalp EEG, and the difficulty of defining what a better fit with the BOLD signal is. When using scalp EEG, it is hard to know the location of the neural activity specifically responsible for, or associated with, the EEG and BOLD signals. We return to the question “What aspects of the electrophysiological
signal best correlate with the amplitude of the co-localised BOLD signal?”, in subsection § 2.2.2, where studies using invasive electrophysiological recordings are reviewed.