Methods in wavelet domain are widely used in various scientific areas as an essential tool in the implementation of filters for satellite, microscopic, telescopic, photographic and medical images, among others. Many of the proposed filters for gray level images are not applicable in MRI, because their implementations do not realize the Rician noise which is the more relevant for this kind of images. This paper describes a wavelet domain filter to remove Rician noise in MRI. The proposed filter uses the Bilateral Filter, the stationary wavelet transform and it is an improvement of those proposed originally by Kazubek. The algorithm is implemented in Matlab and its performance is examined quantitatively by brain images available in Matlab and qualitatively by brain magnetic resonance images of patients with some kind of abnormality.