Standard Echocardiographic View Recognition in Diagnosis of Congenital Heart Defects in Children Using Deep Learning Based on Knowledge Distillation

Abstract

Standard echocardiographic view recognition is a prerequisite for automatic diagnosis of congenital heart defects (CHDs). This study aims to evaluate the feasibility and accuracy of standard echocardiographic view recognition in the diagnosis of CHDs in children using convolutional neural networks (CNNs). A new deep learning-based neural network method was proposed to automatically and efficiently identify commonly used standard echocardiographic views. A total of 367,571 echocardiographic image slices from 3,772 subjects were used to train and validate the proposed echocardiographic view recognition model where 23 standard echocardiographic views commonly used to diagnose CHDs in children were identified. The F1 scores of a majority of views were all ≥0.90, including subcostal sagittal/coronal view of the atrium septum, apical four-chamber view, apical five-chamber view, low parasternal four-chamber view, sax-mid, sax-basal, parasternal long-axis view of the left ventricle (PSLV), suprasternal long-axis view of the entire aortic arch, M-mode echocardiographic recording of the aortic (M-AO) and the left ventricle at the level of the papillary muscle (M-LV), Doppler recording from the mitral valve (DP-MV), the tricuspid valve (DP-TV), the ascending aorta (DP-AAO), the pulmonary valve (DP-PV), and the descending aorta (DP-DAO). This study provides a solid foundation for the subsequent use of artificial intelligence (AI) to identify CHDs in children.

Keywords: congenital heart defect; convolutional neural network; deep learning; knowledge distillation; standard echocardiographic view.

Figure 1

Example images of the 24 standard echocardiographic views: (a) LPS4C; (b) LPS5C; (c) subCAS; (d) subSAS; (e) sub4C; (f) subRVOT; (g) subSALV; (h) DP-MV; (i) DP-PV; (j) DP-DAO; (k) DP-OTHER; (l) DP-TV; (m) DP-AAO; (n) OTHER; (o) A4C; (p) A5C; (q) sax-basal; (r) sax-mid; (s) PSLA; (t) PSPA; (u) supAO; (v) M-OTHER; (w) M-AO; (x) M-LV.

Figure 2

The proposed network architecture for standard echocardiographic view recognition.

Figure 3

The activation maps of the apical four-chamber view and the subcostal sagittal view of the atrium septum. Different colors in the activation map represent different weights in model prediction. The red part has a higher weight and the blue part has a lower weight.

Figure 4

t-SNE visualization of CNN feature clusters for 24 echocardiographic views. Different views are represented with colored clusters and labels. The images are sampled from the test set data and 256 samples were randomly sampled for each view. For views whose total number are <256, all samples are applied.

Figure 5

The confusion matrix between different echocardiographic views.