Clustering of BOLD signals recorded during rest reveals more inter-subject constancy than EEG-fMRI correlation maps

In this paper we apply an hierarchical clustering algorithm to resting state BOLD signals recorded during simultaneous measurement of Electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI). Results of 15 subjects showed that in all cases clusters containing primarily occipital, parietal and frontal lobes were obtained. Furthermore, we found that in all cases, visual and somatosensory cortices were separated into different clusters. Contrary to the inter-subject constancy of the BOLD signal clusters, the statistical parametric maps (SPM’s) resulting from correlating the alpha power time series with BOLD showed a much larger variability, both in terms of spatial distribution and statistical significance. Also the individual EEG spectrograms varied considerably from subject to subject. These results suggest that the BOLD signals have a larger inter-subject constancy and that the variability in the EEG-fMRI SPM’s appears to be due to the variability associated with the EEG alone.