Specified-resolution wavelet analysis of activation patterns from BOLD contrast fMRI

Functional magnetic resonance (MR) MR imagine (fMRI) with blood-oxygenation -level-dependent (BOLD) contras localizes neuronal processing of cognitive paradigms. As magnetic resonance signal responses are small, functional map ping requires statistical analysis of temporally averaged image data. Altho ugh voxels activating at the paradigm frequency can be identified from the Fourier power spectrum, such analyses collapse the temporal information tha t is useful to establish consistency of responses during the paradigm. The design of a set of nonorthogonal wavelets of specified frequency resolution within the power spectrum was investigated for extracting desired frequenc y responses from the noisy signal intensity of individual voxels. These wav elets separate the low-frequency cognitive response to the paradigm from th e respiratory and cardiac responses at higher frequencies. The retention of the temporal information, possible by wavelet analysis, allows the MR sign al changes to be compared to changes in behavioral responses over the durat ion of an entire paradigm. The amplitude and time delay of the wavelet spec ified by the paradigm identify quantitatively the size of the MR signal cha nge and the temporal delay of the hemodynamic BOLD response, respectively. This specified-resolution wavelet analysis was demonstrated for individual voxels and maps through the frontal eye fields using a visually guided sacc ade paradigm.