Identifying the anomaly in LV wall motion using eigenspace

Abstract

In this paper, we have experimented left ventricular (LV) wall motion abnormalities using eigen LV space. We employ three phases of operations in order to perform efficient identification of LV motion abnormalities. In the first phase, LV border detection technique was used to detect LV area. In the second phase, eigen LV spaces of six abnormalities are to be converged as the search space. In the third phase, query is projected on this search space which leads matching of closest disease. The results proved using receiver operating characteristic (ROC) curve show that the proposed architecture provides high contribute to computer-aided diagnosis. Experiments were made on a set of 20 abnormal and 20 normal cases. We trained with eight normal and eight abnormal cases and yielded an accuracy of 88.8% for the proposed works and 75.81% and 79% respectively for earlier works. Our empirical evaluation has a superior diagnosis performance when compared to the performance of other recent works.