Wavelet packet denoising for mammogram enhancement

Abstract

This paper presents an algorithm for denoising and enhancement of mammogram images. The proposed algorithm aims to assist the radiologist toward fast detection and diagnosis of breast cancer. The wavelet packet transform decomposes mammogram images into wavelet packet multiresolution representation. Three different types of mother wavelets have been considered: daubechie-8, Symmlet-8 and Coiflet-5. Nonlinear enhancement function based on the soft-thresholding scheme applied to the decomposed images. A data set of about twenty mammograms with different sizes and types of breast cancer have been used to test the proposed algorithm. The performance of the system is evaluated by two radiologists, who specialized in mammography. Experimental results demonstrate that the proposed technique can be employed to remove noise, enhance, and increase the contrast between malignant and normal tissues.