Detection of Masses in Mammogram Images Using Morphological Operators and Markov Random Fields

In this work we present a two stages method for detection of masses in mammogram images. In a first step, a mass contrast enhancement algorithm based on morphological operators is proposed. Afterwards, a Gaussian Markov Random Field (MRF) model is used for mass segmentation. In the MRF model, the image pixels are described by three statistical texture features of first order. Our approach was tested on 58 mammographic images of masses from MIAS database achieving a performance of 84.4%.