A novel method for benign and malignant characterization of mammographic microcalcifications employing waveatom features and circular complex valued — Extreme Learning Machine
Malar Elangeeran, Savitha Ramasamy, Arumugam Kandaswamy · 2014
This paper presents a novel procedure involving waveatom transform and Circular Complex-valued Extreme Learning Machine (CC-ELM) for automatic characterization of mammographic microcalcifications into benign or malignant. Waveatom transform is used to transform the mammogram image into multi-frequency domain features. The best feature set is obtained by feature reduction through Principal Component Analysis. The reduced feature set is then used to perform classification through a CC-ELM classifier. CC-ELM is a fast learning fully complex-valued classifier to perform real-valued classification tasks efficiently. Mammographic images obtained from Digital Database for Screening Mammography have been used in the study. About 400 Region of Interests extracted from mammograms are used. The performance of the proposed method is about 96.19%, which is significantly higher than the existing methods.