A hybrid of stationary wavelet thresholding and wiener filtering preprocess for noisy brain MRI spatial fuzzy segmentation

This paper proposes a hybrid of stationary wavelet thresholding and Wiener filtering preprocess for spatial fuzzy segmentation of a noisy T1- and T2-weighted MRI. The proposed method reduces the effect of additive Gaussian noise during preprocess. The Gaussian noise, distributed across any wavelet subband coefficients is eliminated by wavelet thresholding technique as well as Wiener filter. Afterwards, fuzzy c-means incorporated with spatial information (sFCM) is utilized for segmenting the denoised image. The denoising preprocess and segmentation measurements rely respectively on peak signal-to-noise ratio (PSNR) and Xie-Beni (XB) validity index. Based on experimental results, the proposed hybrid method improve the segmentation more efficiently than comparative traditional denoising methods.