Paradigm-free mapping with morphological component analysis: getting most out of fMRI data

César Caballero Gaudes, Dimitri Van De Ville, Natalia Petridou, François Lazeyras, Penny Gowland, Manos Papadakis, Dimitri Van De Ville, Vivek K. Goyal
2011 Wavelets and Sparsity XIV  
Functional magnetic resonance imaging (fMRI) is a non-invasive imaging technique that maps the brain's response to neuronal activity based on the blood oxygenation level dependent (BOLD) effect. This work proposes a novel method for fMRI data analysis that enables the decomposition of the fMRI signal in its sources based on morphological descriptors. Beyond traditional fMRI hypothesis-based or blind data-driven exploratory approaches, this method allows the detection of BOLD responses without
more » ... ior timing information. It is based on the deconvolution of the neuronal-related haemodynamic component of the fMRI signal with paradigm free mapping and also furnishes estimates of the movement-related effects, instrumental drifts and physiological fluctuations. Our algorithm is based on an overcomplete representation of the fMRI voxel time series with an additive linear model that is recovered by means of a L 1 -norm regularized least-squares estimators and an adapted block coordinate relaxation procedure. The performance of the technique is evaluated with simulated data and real experimental data acquired at 3T.
doi:10.1117/12.893920 fatcat:pkqdod3trrc5zf4qk4dd5n7gxi