Abstract 15534: Fully Automated "Hands-off" Evaluation of Left Atrial Volume in Children

Abstract

Introduction: Three-dimensional echocardiography (3DE) has been shown to be superior to two-dimensional echocardiography (2DE) for the quantification of chamber volumes and dimensions. However, the utility of 3DE in routine clinical practice has been restricted by time-consuming analysis and the need for 3D-specific training. This study aims to assess the accuracy and clinical utility of Philips Dynamic Heart ModelA.I. (DHM), a “hands off” automated software program, that may be used to measure LA chamber dimensions. This is notable, as indexed LA volume has been used as a predictor of heart failure in adults. Hypothesis: We hypothesize that DHM will be able to quickly and accurately calculate left atrial chamber volume when compared to Cardiac MRI (CMRI) used as a gold standard. Methods: To date, 10 patients (Average Age 14.8±1.9) have undergone clinically indicated CMRI, immediately followed by 3DE, at the Children’s Hospital of Philadelphia. 3D images were analyzed via DHM without manual editing (hands-off approach). DHM automatically calculated maximum LA volume. LA volume was also calculated from CMRI images using the biplane method. LA volume measurements by DHM were correlated with CMRI. Results: DHM was able to calculate LA volume in an average time of 32±8 seconds, without additional human input. Furthermore, inter-technique correlations showed that LA volume measurements made via DHM correlated very well with LA volume measurements made via CMRI (R=0.85, p<0.005). (Figure 1) Conclusions: In children, fully automated evaluation of LA volume based on artificial intelligence is a viable alternative to time-consuming 3D analysis when compared to a CMRI gold standard. This method of “hands-off” analysis may prove to be useful in situations where trained 3D personnel are unavailable. Furthermore, the utilization of this method may result in a greater integration of 3D LA volume into routine clinical practice.