Detection of masses in mammogram images

Breast cancer is the most common cancer in women and is the second leading cause of cancer death. Although it is curable when detected early, about one third of women with breast cancer die, so it is one of the most dangerous types of cancer caused all over the world. In the last decade, many research projects have been carried out aiming to develop computational systems to help specialists in the task of interpreting radiological images. Therefore detection of masses in mammogram images can be used for the early detection of breast cancer. Main contributions of this study are demonstrating the potential of cellular neural networks(CNN) to segment suspect regions in mammographic images and proposing a methodology that includes use of geostatistical functions (Ripley's K function and Moran's and Geary's indices) as texture signatures for mass detection. In the first stage of methodology the image is acquired from the DDSM database which is then pre-processed and later segmented using CNN, further the feature extraction process is carried using geostatistical functions which are later classified using support vector machine. The proposed work can allow this methodology to be added as a computer tool for the medical area, providing support to specialists especially in cases in which visualization is difficult. This allows optimizing the features for higher efficiency.