A good spatial resolution is essential for high precision segmentations of small structures in magnetic resonance images. However, any increase in the spatial resolution results in a decrease of the signal-to-noise ratio (SNR). In this article, this problem is addressed by a new image restoration technique that is used to partly compensate for the loss in SNR. Specifically, a two-stage hybrid image restoration procedure is proposed where the first stage is a Wiener wavelet filter for an initial denoising. The artifacts that will inevitably be produced by this step are subsequently reduced using a recent variant of anisotropic diffusion. The method is applied to magnetic resonance imaging data acquired on a 7-T magnetic resonance imaging scanner and compared with averaged multiple measurements of the same subject. It was found that the effect of image restoration procedure roughly corresponds to averaging across three repeated measurements.
(c) 2010 Wiley-Liss, Inc.