Optimization of mask size for median-modified Wiener filter according to matrix size of computed tomography images

Abstract

We investigated the variation in the noise level and image restoration efficiency as a function of the mask size of the median-modified Wiener filter (MMWF) as applied to computed tomography (CT) images of various matrix sizes. We applied the MMWF with various mask sizes to simulated images of various sizes, which were acquired images using the Shepp–Logan​ phantom and a Male Adult meSH phantom, and we found that for both phantoms, the noise level reduced as the mask size increased. In addition, we calculated the optimized mask size for each matrix size in the case of image-similarity evaluation. Moreover, we conducted an experimental study using the CT image of a chest phantom to evaluate the noise-reduction efficiency of the MMWF with the application of the optimized mask size by calculating the coefficient of variation (COV), signal to noise ratio (SNR), and contrast to noise ratio (CNR). As a result, compared to images without MMWF applied, the COV, SNR, and CNR of image with MMWF were improved by 2.01, 1.02, and 1.60 times, respectively. In conclusion, our findings demonstrate the importance of mask-size optimization (for a given matrix size) in improving the MMWF efficiency. © 2021 Elsevier B.V.