Wavelet-based Mammographic Enhancement

In mammography the interesting characteristics of an image are malignant masses, microcalcifications and skin thickening, of which the last two are said to be indirect signs of malignancy. The mammograms, as normally viewed, display a small percentage of the information they detect and that is due to the minor difference in x-ray attenuation between normal glandular tissues and malignant disease. This makes the detection of small malignancies difficult. The digital medical image processing uses denoising and image enhancement techniques so as to reveal any tumors that may not be obvious and help the oncologist decide. In this paper we employ two different methods of image enhancement and compare them in order to decide which one provides better results in each mammogram. The first method is called adaptive enhancement algorithm and it measures the correlation between wavelet coefficients in successive levels of the analysis so as to decide if the coefficient derives from noise or not and then uses non-linear mapping for the enhancement. The second one, which we call typical method of enhancement, is a method which uses thresholding for the denoising step, non-linear mapping in order to enhance the image and finally filtering to deblur and sharpen the mammogram.