Wavelets and Wavelet Packets for Mammography

In this paper, computer-aided detection of microcalcifications by multiresolution is performed. We firstly present methods of enhancement of mammograms. A combination of these methods is proposed and makes more obvious the features of the image. The second part of this paper deals with the multiscale image processing for detection of microcalcifications by wavelet and wavelet packets. For the wavelet transform, the image is first decomposed into subbands by undecimated (redundant) wavelet transform, the low-frequency subband is then set to zero followed by an adaptive thresholding. For the wavelet packet transform, the image is decomoposed into packets. An adaptive thresholding is then performed to extract microcalcifications. The experimental results have shown good results and the processed images may be used to assist in mammographic interpretation with the potential to diagnose breast cancer at an earlier stage.