Wave-Atom and Cycle-Spinning-Based Noise Reduction in Mammography Images

Abstract

Image denoising is crucial in medical image processing. Digital mammography depends significantly on de-noising for computer-aided-detection of malignant cells like Microcalcifications. In this work, we proposed an unique hybrid approach to reduce Gaussian noise in digital mammograms by combining the wave-atom translation and cycle spinning methods. Pictures denoised by thresholding of coefficients would produce pseudo-Gibbs events because wave atoms are not translationally invariant. Circular motion is applied to keep away the artefacts. Experimental results clearly establish that the method is effective at filtering out background noise while maintaining the integrity of edges and enhancing picture quality. Mini-Mias pictures with variable quantities of Gaussian Noise are used to evaluate and analyse the performance using peak signal-to-noise ratio and structural similarity index. The provided technique outperforms several current filters in terms of evaluated results of peak signal-to-noise ratio and structural similarity index.