Study Limitations. Both the ICA model and the RWMA detection method are sensitive to the quality of the myocardial contours. To construct a good ICA model, high quality myocardial contours are required. This requires a low inter- and intraobserver variation in the contours (if they are manually drawn), or a low segmentation error (if the contours are segmented automatically). This issue is not specific to the proposed method, but it is inherent to any quantitative regional LV function measurement. In the present study, a binary classification between normal and abnormal motion is proposed. Classification of a specific type of abnormal motion, i.e. hypokinetic, aki- netic and dyskinetic, are not presented yet. As yet, the method therefore only serves as a computer-aided tool to draw the clinician’s attention to the suspected abnormal motion areas in the myocardium; staging of the wall motion abnormality may still be performed visually. The current automated method works by modeling contractility patterns for each ven- tricular slice level. Therefore the method does not capture the three dimensional heart
Appears in 2 contracts