An Efficient Method for Detection of Masses in Mammogram Images

Abstract. Breast cancer is one of the most common cancers among women. Mammography is currently the most effective method for early detection of breast cancer. In this paper, a method is proposed for detecting masses in mammogram images. First, based on a specific algorithm, image is segmented and a number of the suspicious regions are obtained. Then, many features are extracted from these regions. To reduce the features, a supervised feature selection method is used. In the final step, a cost-sensitive classifier has been used for classification of the samples. This approach was tested on all images having mass from mini-MIAS data set. Based on the classification results, the percentage of true positive detection rate was 91% false-positive detection was 14% and the area under ROC curve was achieved 96%.