基于DSP的心肺音分離系統(tǒng)研究

陳茜茹 鐘清華 張涵 黃偉新

關(guān)鍵詞： 臨床聽(tīng)診; DSP; 心音; 肺音; 分離系統(tǒng); 小波變換

中圖分類號(hào)： TN945+.5?34; TP391.4 ? ? ? ? ? ? 文獻(xiàn)標(biāo)識(shí)碼： A ? ? ? ? ? ? ? ? ? 文章編號(hào)： 1004?373X（2019）06?0092?04

Abstract： In allusion to the problem that the heart sounds and lung sounds are mixed and also overlapped in the spectrum during the clinical auscultation process， a heart and lung sound separation system based on DSP is designed. The heart and lung sound sensor collects physiological sound signals from the chest wall， which are sent to the signal acquisition and playback circuit for amplification， filtering and digitalization. In the DSP processor， the iterative multi?resolution decomposition and multi?resolution reconstruction （MRD?MRR） separation algorithm based on the wavelet transform is used to separate the heart and lung signals. The separated heart and lung sounds are output by means of audios， and transmitted to the computer via the serial port for backuping or further processing. On the basis of the platform， a heart and lung sound separation experiment was carried out by using the physiological sound signals collected from the chest walls of 30 healthy adults. The results show that the separation algorithm has the best processing effect when using the db8 wavelet function and hard threshold， and the system has a good real?time performance， stable and reliable data transmission， and a high application value.

Keywords： clinical auscultation; DSP; heart sound; lung sound; separation system; wavelet transform

胸部聽(tīng)診是現(xiàn)代醫(yī)學(xué)中診斷心臟系統(tǒng)疾病和呼吸系統(tǒng)疾病的一種重要手段。傳統(tǒng)的聽(tīng)診器由共振片、聲音共振腔和傳導(dǎo)聲音的導(dǎo)管組成，是廣大醫(yī)師最常用的診斷工具。然而它有一些不可彌補(bǔ)的缺點(diǎn)[1]：難以捕捉到人體內(nèi)臟發(fā)出的一些微弱卻非常重要的生理聲音，準(zhǔn)確性差;聽(tīng)診結(jié)果容易受人耳聽(tīng)力敏感局限和聽(tīng)診者主觀經(jīng)驗(yàn)的影響;胸壁聽(tīng)診時(shí)通常為心肺音混合聲音，需要聽(tīng)診者根據(jù)主觀經(jīng)驗(yàn)區(qū)分心音和肺音。這些缺點(diǎn)最終導(dǎo)致診斷過(guò)程極度依賴醫(yī)生主觀經(jīng)驗(yàn)，易造成誤診，錯(cuò)過(guò)患者的最佳治療時(shí)機(jī)。隨著電子技術(shù)和信號(hào)處理技術(shù)的不斷發(fā)展，結(jié)合了現(xiàn)代電子技術(shù)與生物醫(yī)學(xué)技術(shù)的電子聽(tīng)診器研發(fā)成功。電子聽(tīng)診器可以將采集后的信號(hào)進(jìn)行放大等一系列模擬或數(shù)字信號(hào)處理，在一定程度上改良了傳統(tǒng)聽(tīng)診器，但從胸壁檢測(cè)到的生理聲信號(hào)混合了頻譜上相互重疊的心音信號(hào)和肺音信號(hào)，如何將心肺音混合信號(hào)分離出心音信號(hào)和肺音信號(hào)是電子聽(tīng)診器需要解決的重要問(wèn)題[2]。本文介紹了基于DSP的心肺音分離系統(tǒng)和迭代多分辨率分解?多分辨率重構(gòu)（Multi?resolution Decomposition and Multi?resolution Reconstruction，MRD?MRR）分離算法，并通過(guò)實(shí)際測(cè)試，效果很好。

1 ?系統(tǒng)設(shè)計(jì)方案

基于DSP的心肺音分離系統(tǒng)結(jié)構(gòu)如圖1所示，信號(hào)采集部分由心肺音傳感器采集數(shù)據(jù)，并對(duì)心肺音信號(hào)進(jìn)行前期處理，將采集到的模擬信號(hào)轉(zhuǎn)化為數(shù)字信號(hào);DSP 處理器對(duì)心音肺音信號(hào)進(jìn)行分離等處理以及對(duì)相關(guān)外部設(shè)備的控制;存儲(chǔ)部分主要是擴(kuò)展的程序存儲(chǔ)和數(shù)據(jù)存儲(chǔ)，由FPGA，F(xiàn)LASH 和SDRAM 組成。另外，DSP 還可通過(guò)USB 接口和PC進(jìn)行通信，也可直接和SDCARD 進(jìn)行通信。心肺音分離算法由DSP軟件實(shí)現(xiàn)，分離后的心音和肺音一方面通過(guò)音頻輸出，另一方面通過(guò)串口傳輸?shù)接?jì)算機(jī)中備份或進(jìn)行進(jìn)一步處理。