Denoising and Enhancing Image Quality for Detection Accuracy in Mammograms

Abstract

Since mammography was introduced in the middle of the 20th century, it has been the preferred method for detecting breast cancer, thereby significantly impacting the prognosis and survival rates of affected individuals. However, the diagnostic accuracy of mammographic images can be compromised by the presence of noise, low contrast, indistinct features, and poor differentiation from surrounding tissue. This paper presents a comprehensive approach that addresses the limitations of mammograms by combining wavelet-based denoising and morphological transformation for denoising and enhancing mammographic images. This work reduces the impact of noise artifacts while preserving image features by applying various shrinkage thresholds to the subbands of the stationary wavelet transform. Subsequently, a novel morphological filtering operation is applied to further enhance contrast, suppress noise, refine edges, and remove unwanted artifacts, all while emphasizing relevant structures. The experimental results outperformed existing methods in terms of quantitative metrics such as PSNR, SSIM and MSE. Notably, it also achieved superior performance compared to the common enhancing method, CLAHE. This proposed method is a promising new approach for denoising and enhancing mammograms. It has the potential to contribute to earlier and more accurate breast cancer diagnosis, thereby advancing the fields of women's health and cancer detection.