DETECTION OF MICRO-CALCIFICATION IN MAMMOGRAPHIC IMAGES USING ARTIFICIAL NEURAL NETWORK METHOD BASED ON STATISTICAL FEATURES

摘要

The detection of micro calcification cluster (MCCs) during mammographie examination is vital for the correct diagnosis of breast cancer in asymptomatic women. Automated detection methods augment the diagnostic process by providing the radiologist with a second opinion, in this paper we present a novel approach to the problem of computer aided analysis of digital mammograms for the breast cancer detection. This paper takes digital mammography for scrutiny with the angle of finding the micro-calcification in the mammographie images through the help of artificial neural networks (ANN) and wavelet-based sub band image decomposition. When the mammographs are digitized the micro calcification present in it will be in the form of high frequency component of the image matrix. In order to detect it we filter the image using Hessian filter and apply DWT and finding the Skewness and Kurtosis of the resulting image, before applying the BPN algorithm for diagnosing the cancer. The neural network contains one input, two hidden and one output layers. He proposed methodology is tested using the Nijmegen and the Mammographie Image Analysis Society (MIAS) Mammographie databases.