Digital Signal Processing 17 (2007) 711–723 www.elsevier.com/locate/dsp Impulse noise reduction in medical images with the use of switch mode fuzzy adaptive median filter Abdullah Toprak a , ˙ Inan Güler b,∗ a Dicle University, Meslek Yüksek Okulu, Elektrik-Elektronik Bölümü, 21280 Diyarbakır, Turkey b Gazi University, Teknik E˘ gitim Fakültesi, Elektronik-Bilgisayar Bölümü, 06500 Teknikokullar, Ankara, Turkey Available online 15 December 2006 Abstract In this paper, a novel fuzzy adaptive median filter is presented for the noise reduction in MR images corrupted with heavy impulse (salt&pepper) noise. We propose a switch mode fuzzy adaptive median filter (SMFAMF) for removing highly corrupted salt&pepper noise without destroying edges and details in the image. The SMFAMF filter is an improved version of adaptive median filter (AMF) in order to reduce additive impulse noise in the images. The proposed filter can preserve details in the images better than AMF while suppressing additive salt&pepper or impulse type noises. In this paper, we placed our preference on bell- shaped membership function with adaptive parameters instead of triangular membership function without variable coefficients in order to observe better results. Experiments with the magnetic resonance (MR) image from healthy subject, an MR image having the opaque material, and an MR image having disease demonstrate the mean square error (MSE), root mean square error (RMSE), signal-to-noise ratio (SNR), and peak signal-to-noise ratio (PSNR) of the proposed method. The results show that the proposed method can be useful for MR images with impulse type noises.  2006 Elsevier Inc. All rights reserved. Keywords: Adaptive median filter; Fuzzy adaptive median filter; Impulse noise; Noise reduction 1. Introduction Medical images are often deteriorated by noise due to various sources of interference and other phenomena that affect the measurement processes in imaging and data acquisition systems. Median filtering is a common nonlinear method for noise suppression that has unique characteristics. It does not use convolution to process the image with a kernel of coefficients. Rather, in each position of the kernel frame, a pixel of the input image contained in the frame is selected to become the output pixel located at the coordinates of the kernel center. The kernel frame is centered on each pixel (m, n) of the original image, and the median value of the pixels within the kernel frame is computed. The pixel at the coordinates (m, n) of the output image is set to this median value [1]. Median filter (MF) is a 2D image filter that is more effective in situations where the impulse noise is less than 0.2 [2]. If this ratio exceeds 0.2, adaptive median filter (AMF) is used. As it is the case in the other filters, an S xy window is selected for the AMF [3]. However, a feature that differentiates the AMF from the other filters is the fact * Corresponding author. E-mail addresses: atoprak@dicle.edu.tr (A. Toprak), iguler@gazi.edu.tr ( ˙ I. Güler). 1051-2004/$ – see front matter  2006 Elsevier Inc. All rights reserved. doi:10.1016/j.dsp.2006.11.008