Abstract

Info

Author(s):: LI Jiang; LI Piding; School of Health Science and Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China

Keywords:: Keywords: congenital heart disease screening system convolutional neural network wavelet denoising second-order spectrum

Abstract:: Abstract: A congenital heart disease screening system based on convolutional neural network is designed for improving the efficiency of doctors in screening for congenital heart disease. The system realizes the real-time synchronous acquisition, visualization and quantitative analysis of physiological parameters such as heart sounds and ECG data with the cooperation of software and hardware. The system includes upper and lower computers. The lower computer uses FPGA as the core to complete the preprocessing such as heart sound and ECG data acquisition and wavelet threshold denoising, while the upper computer implements the second-order spectral feature extraction, the binary classification using convolutional neural network, and the visual display on the user interface in the Python programming language under the Windows environment. The established screening system is tested on 200 volunteers, and finally achieves an accuracy rate of 94.5%, a specificity of 95.9%, and a sensitivity of 93.2%, indicating that the system has good performance and can provide effective assistance for clinical diagnosis of congenital heart disease.