Classification of Cardiac Ultrasound Image Sequences Based on Sparse Representation

摘要

To classify thrombosis and pectinate muscle incardiac ultrasound image sequences, a classification methodbased on sparse representation is proposed. This methodextracts GLCM based texture features to form the sampleset and compute the sparse solution with coefficients how atest sample be represented by the training set. After that,two kinds of constraints and classification strategy areadded to achieve the classification. Experiment resultsshows that the proposed approach can achieve aclassification accuracy of 91.92%, significantly higher thanother popular classifiers.