In Journal of the American Society of Echocardiography : official publication of the American Society of Echocardiography
BACKGROUND : Echocardiography-based screening for valvular disease in at-risk asymptomatic children can result in early diagnosis. These screening programs, however, are resource intensive, and may not be feasible in many resource-limited settings. Automated echocardiographic diagnosis may enable more widespread echocardiographic screening, early diagnosis, and improved outcomes. In this feasibility study, we sought to build a machine learning model capable of identifying mitral regurgitation (MR) on echocardiogram.
METHODS : Echocardiograms were labeled by clip for view and by frame for the presence of MR. The labeled data were used to build two convolutional neural networks (CNNs) to perform the stepwise tasks of classifying the clips 1) by view and 2) by the presence of any MR, including physiologic, in parasternal long axis color Doppler views (PLAX-C). We developed the view classification model using 66,330 frames and evaluated model performance using a hold-out testing dataset with 45 echocardiograms (11,730 frames). We developed the MR detection model using 938 frames and evaluated model performance using a hold-out testing dataset with 42 echocardiograms (182 frames). Metrics to evaluate model performance included accuracy, precision, recall, F1 score (average of precision and recall, 0 to 1 with 1 suggesting perfect precision and recall), and receiver-operating characteristic analysis.
RESULTS : For the PLAX-C view, the view classification CNN achieved an F1 score of 0.97. The MR detection CNN achieved a testing accuracy of 0.86 and an area under the receiver operating characteristic curve of 0.91.
CONCLUSIONS : A machine learning model is capable of discerning MR on transthoracic echocardiography. This is an encouraging step toward machine learning-based diagnosis of valvular heart disease on pediatric echocardiograms.
Edwards Lindsay A, Feng Fei, Iqbal Mehreen, Fu Yong, Sanyahumbi Amy, Hao Shiying, McElhinnney Doff B, Ling X Bruce, Sable Craig, Luo Jiajia
Machine learning, deep learning, echocardiogram, mitral valve regurgitation