Mass detection algorithm based on support vector machine and relevance feedback

Abstract

To improve the detection of mass with appearance that borders on the similarity between mass and density tissues in the breast, an support vector machine classifier based on typical features is designed to classify the region of interest (ROI). Furthermore, relevance feedback is introduced to improve the performance of support vector machines. A new mass detection scheme based on the support vector machine and the relevance feedback is proposed. Simulation experiments on mammograms illustrate that the novel support vector machine classifier based on typical features can improve the detection performance of the featureless classifier by 5%, while the introduction of relevance feedback can further improve the detection performance to about 90%.

Keywords

support vector machine / relevance feedback / mass detection / feature extraction