There is recent evidence that functional magnetic resonance images (fMRI) have a 1/w-like spectrum which is characteristic of self-similar processes. This finding is also supported by the factal-like structure of the brain vasculature (cf. enclosed image). The main purpose of this project is to take advantage of this property and to develop an optimal Wiener filtering algorithm for the reduction of noise in fMRI images of the brain. This Wiener filter will be specified by a small number of parameters (typically, the spectral slope and the SNR) which should eventually be estimated from the data. The algorithm will be implemented in the Fourier domain as a Plug-in for ImageJ so that it can be made available over the WEB.