Implementation of a Selective Median Filter in Computed Tomography for Image Quality Enhancement

Abstract

Noise in computed tomography (CT) images is unavoidable due to various factors such as errors from patient sources, hardware errors, and image acquisition. In order to reduce the noise, a selective median filter (SMeF) has been developed. SMeF determine the median value selectively using a threshold. For quantitative evaluation, effectiveness of the filter algorithm was assessed using metrics of modulation transfer function (MTF) and contrast-to-noise ratio (CNR) obtained from phantom and head CT images. The results of the SMeF assessment were compared with the original image and images filtered with traditional median filter (MeF), selective mean filter (SMF) and bilateral filter (BF). It was found that the noise levels of images filtered with SMeF are lower compared to those filtered with MeF, BF, and SMF. SMeF has a relatively good value in terms of CNR. The spatial resolution of images filtered with SMeF is comparable compared spatial resolutions of the original image, MeF, SMF and BF. The SMeF is an effective algorithm in reducing noise of CT phantom images and clinical images.