Ischemia disease recognition in consecutive stress echocardiography using sparse representation based classifier

In this paper, recognition of the abnormal motion of the left ventricle in consecutive echocardiography frames is evaluated for ischemia detection. For this purpose, a classifier is designed based on the sparse representation concept. This classifier is constructed based on dictionary construction for inside and outside of the left ventricle. Two groups of dictionaries are defined based on feature vectors obtained by extracted features with the use of grey level co-occurrence matrix. The left ventricular wall motion is analyzed and a determinant parameter for the abnormal motion detection would be obtained based on this movement. This parameter is normalized with left ventricle long axis and the heart rates in which the stress echo test has been taken. The obtained plot for this parameter in different heart rates shows difference in the slope and the amount of maximum variation in different cases. This parameter can be used as a criterion for separation of the cases with ischemia and the normal ones

Keywords:

Feature extraction , GLCM , Sparse representation-based classifier , Stress echocardiography , Left ventricle wall motion