Real-time scan assistant for echocardiography

Abstract

A real-time scan assistant (SA) for use with echocardiography is presented. The motivation is to aid nonexpert users in capturing apical 4-chamber views (A4CH) during echocardiography. The algorithm is based on a parametric multi-chamber model of the A4CH view, updated in an extended Kalman filter framework. The regional model goodness-of-fit is used to calculate a score, which is provided to the user during acquisition, together with an icon (emoticon) indicating whether the current view is acceptable or not. The SA was implemented on a commercially available scanner. A feasibility test was performed using two healthy volunteers as models and 10 medical students acting as nonexpert users. The students examined the models on two occasions, separated more than four days in time. Half of the students used the SA during the first exam and no SA at the second exam. The other half used the opposite order. The recordings were later rated by a cardiologist. A Wilcoxon signed pair rank test revealed a statistically significant improvement when using SA. Nine cases were rated as poor without using the SA. In eight (89%) of these cases, view quality improved to acceptable when the SA was used.