Optimization of factor analysis of the left ventricle in echocardiography for detecting wall motion abnormalities

This study has tested solutions to optimize the ability of factor analysis of the left ventricle in echocardiography (FALVE) to detect segmental wall motion abnormalities automatically. On four- and two-chamber views of 38 patients, two factors (one flat curve and one curve describing the contraction-relaxation sequence) were extracted and associated factor images were combined to synthesize a parametric image (constant image in green, positive/negative values of the contraction-relaxation image in red/blue). The segments were graded on the visual and the parametric views. The impact of selecting a whole cardiac cycle, masking the left ventricle and realigning the image sequence on the results, was demonstrated. Systematic realignment had a positive impact, especially for patients with left bundle branch block or pacemaker. After alignment, for the entire population, the absolute concordance was 68.6% and the relative concordance (within one grade) was 99%. Thus, FALVE is promising for detecting segmental wall motion abnormalities. (E-mail: benoit.diebold@egp.ap-hop-paris.fr)