Group Independent Component Analysis for mining concurrent EEG-fMRI data

Eichele, T.
Biologische und Medizinische Psychologie, Universität Bergen

Independent component analysis (ICA) is a data-driven multivariate approach that has become increasingly popular for analyzing brain imaging data. ICA allows extraction of temporally or spatially independent source signals that constitute the observed data without spatial and temporal priors. In this presentation, I will briefly introduce the rationale of group-level extensions of spatial ICA for hemodynamic (fMRI) and temporal ICA for electrophysiological (EEG) analysis, and then lay out how group models can be used for integration of multidimensional data spaces such as concurrent EEG-fMRI. In particular, I will describe symmetric fusion with parallel and joint models, partly employing deconvolution of hemodynamic responses and single trial estimation to assess the coupling between modalities using examples from event-related and resting state studies. Considering the limitations of decompositions with ICA, I will discuss some of the outstanding challenges for multimodal fusion methods.

Symposium 20: Recent Advances in EEG-fMRI Integration
12.06.2009, 14:00-15:00
Seminarraum 11