So far, the most recognized software to analyze data from fMRI
experiments is Statistical Parametric Mapping (SPM). But 6 years
ago an alternative approach based on spatial wavelet transform
instead of Gaussian pre-filtering (as it is the case with SPM)
has been proposed. It is based on the sparsity provided by the
representation in the wavelet domain that increases the SNR as
the noise remains evenly distributed in this domain.
We describe a scientific work on a new manner of exploiting fMRI
data based on the latest idea of introducing redundant wavelet
transforms. Redundant wavelet processing and frame theories more
generally suggested hopes of achieving even finer results than
with classical non-redundant wavelet.