Enhancement of mammogram by hyper elastic property of non rigid Images: A histogram modification scheme [articol]

Abstract

Mammogram enhancement is of paramount importance for the successful detection of masses and micro calcifications. We propose a hyper elastic model based new histogram modification scheme that mimics the deformation property of the breast. Unlike state-of-the-art methods, the advantage of using such model is that the anatomical topology of masses and micro calcifications is preserved. Our approach is fully tested on the public available mini-MIAS database. The experimental results has earned higher average value of 7.57 in terms of effective measure of enhancement which clearly outperforms traditional techniques such as histogram equalization (1.13), contrast limited adaptive histogram equalization (6.37),brightness preserving bi-histogram equalization (2.34) and recursive mean separate histogram equalization(6.47). Most significantly, the visual inspection indicates that our method preserves the topological information of malignant breast structures without eliminating relevant image features.