一种新生儿先心病筛查仪器的设计  

Design of a Newborn Congenital Heart Disease Screening Instrument

作　　者：崔昊[1] 李丕丁[1] 

机构地区：[1]上海理工大学健康科学与工程学院,上海

出　　处：《应用物理》2023年第11期439-452,共14页Applied Physics

摘　　要：先天性心脏病是我国最常见的先天性缺陷。为了尽早筛查诊断和治疗,以降低新生儿死亡率,本文设计了一款多功能的专用于新生儿先心病的筛查仪器。该仪器集信号采集、通信、处理和分析为一体,包括四组高性能的电子听诊器、三导联心电、血氧传感器和FPGA嵌入式系统等。它能够同步采集四个主要听诊区的心音,并结合心电信号进行心音分割。采用MFCC提取心音信号特征,最终通过卷积神经网络进行分类识别。实验结果表明,本文设计的仪器准确率为达96.7%,对异常心音具有较高的识别率,可以辅助医护人员进行先心病筛查工作。Congenital heart disease is the most common congenital defect in China. In order to screen, diagnose, and treat it as early as possible to reduce neonatal mortality, a multifunctional screening device specifically designed for newborns with congenital heart disease is designed in this study. This instrument integrates signal acquisition, communication, processing, and analysis, includ-ing four groups of high-performance electronic stethoscopes, a three-lead electrocardiogram, a blood oxygen sensor, and an FPGA embedded system. It can synchronously collect heart sounds from four major auscultation areas and combine them with electrocardiographic signals for heart sound segmentation. The MFCC technique is employed to extract features from heart sound signals, and the classification and recognition are conducted through a convolutional neural network. The experimental results show that the accuracy of the instrument designed in this paper is 96.7%, which has a high recognition rate for abnormal heart sounds and can assist medical staff in the screening of congenital heart disease.

关 键 词：先天性心脏病 心音 医疗器械 

分 类 号：TN9[电子电信—信息与通信工程]