We investigated an integration strategy whereby relevant EEG features are extracted and linear models learnt so as to predict the Default Mode Network activity measured by simultaneous fMRI. We compared the performance of four models: root-mean-squared frequency (RMSF), total power (TP), linear combination of band-specific power (LC), and weighted degree of the functional connectivity network built from the band-specific imaginary part of coherency of the EEG cross-spectrum (WD-ICoh). Models were estimated using elastic net regularization and were found to predict the target BOLD signal with fairly good correlations. Although these varied significantly across models, WD-ICoh outperformed the remaining models.
This abstract and the presentation materials are available to members only; a login is required.