Smoothing filters in synthetic cerebral magnetic resonance images: A comparative study

Abstract

This paper presents the evaluation of two computationaltechniques for smoothing noise that might be presentin synthetic images or numerical phantoms of magneticresonance (MRI). The images that will serve as the databases(DB) during the course of this evaluation are availablefreely on the Internet and are reported in specializedliterature as synthetic images called BrainWeb. Theimages that belong to this DB were contaminated withRician noise, this being the most frequent type of noisein real MRI images. Also, the techniques that are usuallyconsidered to minimize the impact of Rician noise on thequality of BrainWeb images are matched with the Gaussianfilter (GF) and an anisotropic diffusion filter, based onthe gradient of the image (GADF). Each of these filters has2 parameters that control their operation and, therefore,undergo a rigorous tuning process to identify the optimalvalues that guarantee the best performance of both theGF and the GADF. The peak of the signal-to-noise ratio(PSNR) and the computation time are considered as keyelements to analyze the behavior of each of the filteringtechniques applied. The results indicate that: a) both filtersgenerate PSNR values comparable to each other. b)The GF requires a significantly shorter computation timeto soften the Rician noise present in the considered DB.