基于电子听诊器的心音分类系统

Heart Sound Classification System Based on Electronic Stethoscope

为解决传统听诊器的局限性,设计基于STM32F103单片机的新型电子听诊器,并设计心音分类软件,有利于临床听诊。医疗工作者可使用3.5mm耳机或者蓝牙播放设备进行听诊,并设置0~10dB的音量增益。通过对数字滤波器截止频率的设定,该听诊器分3种听诊模式,可实现全频带、心音和肺音听诊。心音数据通过TCP/IP协议传输到心音分类软件,构成局域网系统。通过决策树模型实现心音的异常判别,准确率高达91.24%,灵敏度98%。该听诊器的听诊音清晰易识、佩戴感舒适、功耗低,可实现数字化听诊及心音的异常判别。

In order to solve the limitations of traditional stethoscope, this paper designs a new electronic stethoscope based on STM32F103, and designs heart sound classification software, which is conducive to clinical auscultation. The physician can use a 3.5mm headset or Bluetooth playback device for auscultation and set a volume gain of 0~10dB. By setting the cutoff frequency of the digital filter, the stethoscope can be divided into three auscultation modes to achieve full-band, heart sound and lung sound auscultation. The heart sound data is transmitted to the heart sound classification software through the TCP/IP protocol to form a LAN system. The decision tree model is used to realize the abnormality of heart sound, and the accuracy rate is up to 91.24% and the sensitivity is 98%. The auscultation sound of the stethoscope is clear and easy to understand, comfortable to wear, and low in power consumption, and can realize digital auscultation and abnormal discrimination of heart sound.