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

Lanping Wu,Xiaoqing Liu,Qiuyang Sheng,Yizhou Yu,Wenjing Hong,Yuqi Zhang,Lijun Chen,Bin Dong,Liebin Zhao,Kunlun Gao

doi:10.3389/fped.2021.770182

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.

Highlights

Congenital heart defects (CHDs) are the most common birth defects in China, with an incidence rate of approximately 0.9% in live births, and are the main causes of death in children aged 0–5 years [1]
Six professional ultrasound doctors participated in the data annotation, of which four ultrasound doctors participated in the annotation of the training data set, and two ultrasound doctors participated in the annotation of the test data set
We will add more training data to improve the performance of the proposed model

Summary

Introduction

Congenital heart defects (CHDs) are the most common birth defects in China, with an incidence rate of approximately 0.9% in live births, and are the main causes of death in children aged 0–5 years [1]. Precise preoperative diagnosis helps to develop the most reasonable surgery and interventional treatment plan for CHDs. Transthoracic echocardiography (TTE) can provide sufficient information about the structure of the heart to evaluate hemodynamics and cardiac. It is currently the most commonly used non-invasive examination method for CHDs. Many factors contribute to the heterogeneity of echocardiographic diagnosis, such as inherent pulse variability in the heart, speckled noise and artifacts in echocardiogram, and inter- and intra-class differences in standard echocardiographic views [2]. The anatomical structure and spatial configuration of CHDs are complex and changeable, and accurate diagnosis through TTE is complicated and timeconsuming, and it is heavily dependent on the accurate judgment of each echocardiographic view by experienced cardiologists. In China, there is a lack of experienced cardiologists at the grassroots level, especially in rural areas. It is very necessary to establish an automatic diagnosis system of CHDs to alleviate the difficulty of CHD diagnosis at the grassroots level

Objectives

Methods

Discussion

Conclusion