Analysis Methods for fMRI Data Sets
This project deals with the development of new signal analysis methods for fMRI data sets. The size of fMRI data sets make the full application of standard corelational techniques intractable. Wavelets transforms are being explored as a means for both denoising and detrending data, as well as for data reduction, with the goal of applying dense correlational techniques in the wavelet domain where these may be more tractable. A second focus is on Independent Component Analysis (lCA), which has been proposed as a method for "blind source separation" of generators in fMRI data sets. Other approaches are being explored as well, which would integrate temporal and spatial information, permitting the use of the multiresolution techniques, within which to apply ICA or other blind source separation approaches.
Contact:
Ingrid Daubechies