Mammograms provide a useful tool for diagnosing breast cancer. It is very difficult to classify the microcalcifications as benign or malignant directly by the radiologist from the mammogram images. Therefore in this paper a comparative study based on the curvelet and waveatom features extracted from the mammograms and classification using various classifiers such as Naïve Bayes, ELM (Extreme Learning Machine) and complex ELM has been presented which can be used as a CAD (Computer Aided Diagnosis) system for microcalcification detection. The experimental results were obtained by training and testing data with different classifiers and were compared using classification accuracy obtained. From the results, it was found that the complex extreme learning machine was the best classifier for the waveatom features