以呼吸聲頻之數位訊號作哮喘病徵之辨識

The use of the digital breathing frequency signal for wheezing recognition

陳文哲(Wen-Che Chen)

指導教授 ： 謝傳璋

國立臺灣大學/工學院/工程科學及海洋工程學研究所/碩士(2011年)

https://doi.org/10.6342/NTU.2011.00280

摘要

氣喘，又稱為哮喘，是一種慢性支氣管發炎的病症。一般常以聽診器經過醫師等專業的醫護人員聽取肺音作診斷，就肺音而言，常以哮喘音為典型的氣喘發作特徵。台灣醫護人員的人力不足以及看護工或家庭照顧者並沒有氣喘方面的專業知識，因此本研究的目的在於建立一套可以應用於居家照護或長時間監測氣喘發作的小型裝置。
本研究藉由工研院所開發之軟性駐極體裝置，經由頸部擷取肺音訊號，利用數位訊號處理之時頻分析，偵測出異常的肺音，進而判斷哮喘發作之特徵。在時頻分析方面，本研究採用了希爾伯特黃轉換(Hilbert-Huang Transform, HHT)、短時傅立葉轉換(Short-Time Fourier Transform, STFT)及自相關函數(Auto Correlation Function, ACF)，三種方法去分析並比較其結果。
本論文主要以電腦呼吸音分析(Computerized Respiratory Sound Analysis, CORSA)定義哮喘音之標準為基礎，下載網路資料庫上的肺音訊號作分析，正確率分別為希爾伯特黃轉換法94.83%，自相關函數法93.1%，短時傅立葉轉換法91.38%；最後由台大醫院胸腔門診的氣喘病患量測的臨床訊號作驗證，辨識率之結果分別為希爾伯特黃轉換法85%，自相關函數法70%，短時傅立葉轉換法78%。因此，本研究選擇HHT方法作為哮喘音的辨識，期許未來可以應用在小型嵌入式系統作居家照護與長時氣喘病患的監控上。

關鍵字

肺音 ； 哮喘音 ； 數位訊號處理 ； HHT ； STFT ； ACF

並列摘要

Asthma is a common chronic inflammatory disease. Through the stethoscope the characteristics of lung sounds are taken for diagnosis by physicians usually. Wheezing is a typical feature of asthma attack.
The purpose of this study is to establish a small device for asthma attack monitoring in long-term care.
In this study, three methods, say , the Hilbert-Huang Transform(HHT), the Auto Correlation Function(ACF) and the Short Time Fourier Transform(STFT) were used to detect the threshold of Wheezing for medical alarm.
The criterion suggested by the Computerized Respiratory Sound Analysis (CORSA) for wheezing detection is adopted in this study. By using the sample data down load from the data base, the experimental result shows that the identification rates are 94.83% for HHT, 93.1% for ACF and 91.38% for STFT respectively. On the other hand, the lung sound signal of asthma patient measured in the National Taiwan University Hospital chest clinic are used for validation of these three methods. The identification rates are 85% for HHT, 70% for ACF and 78% for STFT respectively. Therefore, this study shows that the HHT method is a better choice for wheezing recognition.
In the future, it is hope that this algorithm can be used in the home-care asthma monitoring system.

並列關鍵字

Lung sound ； HHT ； ACF ； STFT ； Wheeze

參考文獻

10.楊佳穎, 以HHT為基礎之肺音分析與哮喘音辨識研究. 國立台北科技大學, 2008. 碩士論文.

4.M. Mahagnah, a.N.G., Repeatability of Measurement of Normal Lung Sound. Am. J. Respir. Crit. Care Med., 1994. vol. 149: p. 477-481.

5.Y. Shabtai Musih, B.G.J., and G. Noam, Spectral content of forced expiratory wheezes during air, He, and SF6 breathing in normal humans,. journal of applied Physiology, 1992. vol. 72(no. 2): p. 629-635.

7.A. Homs Corbera, R.J., J. A. Fiz, and J. Morera, Algorithm for time-frequency detection and analysis of wheezes. proceedings of the 22nd IEEE Annual International Conference on Engineering in Medicine and Biology Society, 2000. vol. 4: p. 2977-2980.

8.R. J. Riella, P.N., R. F. Borges, and A. L. Stelle, Automatic wheezing recognition in recorded lung sounds. proceedings of the 25nd IEEE Annual International Conference on Engineering in Medicine and Biology Society, 2003. vol. 3: p. 2535-2538.