Wavelet methods are increasingly being used to facilitate signal extraction and estimation in many applications. This is especially true in neuroimaging where large spatio-temporal data sets containing correlated noise structure require sensitive methods to detect the underlying signal.

This talk will briefly introduce wavelets and their application to Positron Emission Tomography and functional Magnetic Resonance Images. It will be demonstrated that linear temporal models can be routinely applied to the wavelet coefficients as opposed to the image coefficients, to derive temporal parameters of interest. Denoising routines for these parameters will also be explored.

Neuroimaging parameters derived in image space are also routinely accompanied by their corresponding error estimates. In order for wavelet methods to be more generally used in neuroimaging, these error estimates are also required when using wavelet methods. A new method for the derivation of the image space error estimates from their corresponding wavelet errors under certain assumptions will be presented. These can be calculated using rapid algorithms which reduce the computational burden of this transformation.