Performance evaluation of median-modified Wiener filter algorithm in high-resolution complementary metal–oxide–semiconductor radio-magnetic X-ray imaging system: An experimental study

Abstract

Noise reduction is a fundamental early stage in X-ray image performance evaluation. In this study, the median-modified Wiener filter (MMWF) algorithm based on a nonlinear adaptive spatial filter is compared to two general noise-reduction techniques using median and Wiener filters to confirm an excellent denoising approach in X-ray images. To acquire images, a high-resolution complementary metal–oxide–semiconductor (CMOS) radio-magnetic X-ray imaging system (exposure conditions: 100, 400, and 700μA) and rat phantom were used. The performances of the denoising methods were evaluated in terms of contrast-to-noise ratio (CNR), coefficient of variation (COV), and no-reference image quality assessment using the Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE). The average results for CNR, COV, and BRISQUE in the acquired X-ray image using the MMWF algorithm were 1.08, 1.10, and 1.03 times higher than those of the noisy image, respectively. On average, the MMWF algorithm provided better image restoration than general noise-reduction techniques and was found to be most effective in relatively lower exposure conditions.