Wavelet based feature extraction method for breast cancer diagnosis
Sarah Benmazou, Hayet Farida Merouani · 2018
Breast cancer is known as one of the important diseases in medical science in which its early detection can reduce mortality rate and improve the survival rate of the patients. A World Health Organization (WHO) estimates that in 2015, at least 561 thousand women will die of breast cancer throughout the world. Early detection of cancer can reduce the rate of mortality, so it would be useful to develop a computer-assisted method to help differentiate between masses or tumors based on characteristics extracted from the part of interest in mammography. In this article, the idea is to combine descriptors of different nature for digital mammography: texture and shape. The texture descriptor (Wavelet), which is often used for the classification of mammary masses and the shape descriptor SMD (Spiculated Mass Descriptor), which is very effective from the point of view of robustness to noise, invariance to geometric transformations and characterization of masses breast.