摘要
肺部聽診是一種簡單，且非侵入式的肺部疾病診斷方法。病患發生肺音異常的狀況，常會因環境與當時的身體狀況有關，由於醫護人員無法長時對病患做肺音的聽診，所以長時監控病患肺音有其必要性。因此本研究發展一套電腦輔助肺音監測系統，提供一個較為客觀且有效的肺音萃取系統。系統可分為類比線路與數位訊號處理兩部分，其中類比線路部分包含四組麥克風線路、放大與濾波電路、ECG量測電路和氣壓感測線路等。並利用16-bit的訊號擷取卡將生理訊號擷取至電腦中。在數位訊號處理端中，研究中嘗試使用一個新的視窗時域分析與ICA兩種方式，將心音與肺音個別萃取出來。視窗時域分析法，主要是利用ECG與呼吸時的氣壓變化，分別偵測心音與肺音發生區段；並利用視窗法將其特定訊號強度減弱，已達到訊號分離的目的。在ICA的處理中，主要利用統計模式找尋三組生理聲音的主成分，作為心音與肺音成分離的方法。
在系統驗證方面，研究中收集3個正常男性受測者（平均年齡在25±3歲），無肺部疾病歷史。分析結果發現，在使用視窗的時域分析下，可以完整的偵測出心音與肺音區間。並在視窗法的效應下，有效降低心音與肺音的互相干擾，達到分離訊號的目的。在ICA分析的結果中，並不能有效分離心音與肺音的訊號；在文獻的探討中推測，可能是胸腔共鳴與擷取的生理聲音不足，而導致無法取得較佳的結果。在實際臨床病患量測方面，本研究實際偵測一名長期使用呼吸器的慢性阻塞性肺疾病（COPD）男性病患，其年齡為62歲。由於該病患配置呼吸器，無法直接使用研究中所設計的氣壓感測器，所以本研究改採用量測呼吸器出氣端的氣流聲，並利用希爾伯特轉換的方式來偵測呼吸區間。實驗結果發現，利用視窗的時域分析可將病患的心音強度抑制，並降低其對肺音的干擾。另外，搭配希爾伯特轉換的呼吸區間偵測，也能對無法直接使用氣壓感測器的臨床病患，得到另一個有效的替代方式。在長時量測與偵測病症特徵方面，這套系統可以穩定長時量測，並可搭配簡單判斷法則，即可對異常肺音做出即時的偵測。此系統未來可運用於即時儲存正常人與長期臥床病患的肺音，並可做出即時偵測特定病徵的需求。

Lung auscultation is a simple, non-invasive and indispensable physical diagnosis of respiratory disease. Abnormal lung sounds are usually relevant to the airway condition of the patients. However, it is difficult to monitor the lung sounds of the patients for a long period of time. It is often necessary and helpful for those who are bedridden, especially in ICU and RCC. The present study developed an acquisition system computer-aided of lung sounds to offer a comparatively objective and effective lung sounds analytic system. This system consist two components: a data collecting hardware and a digital post-processing hardware. The formation compasses a microphone, an amplifier, a filter circuit, an ECG circuit, and a pressure sensor air flow. In this circuit, a 16-bit A/D converter applied to condition the input data. Regarding the digital signal processing, this study used a new windowing analysis base on time-domain or ICA method to separate the raw data comprising heart and lung sounds into two independent signals. Windowing analysis based on time-domain effectively separated the heart and lung sounds with the ECG and air flow signals as references. For the ICA method, a statistic model was adapted to find the main component of lung sounds from the three channels of physiological sounds.
Three healthy males (mean age 25 years) were selected to verify the developed system. The results showed that the windowing method successfully detected and extracted the heart sounds and lung sounds separately. However, the ICA analyses could not separate the signals of heart sounds and lungs sound effectively. This might be due to thorax resonance and insufficient channels of physiological sounds for the ICA analysis. In the clinical experiment, a 62-year-old, chronic obstruction pulmonary disease (COPD) male patient and supports of mechanical ventilator over a long period of time, was examinated with this system. To compensate the ventilator dependence, the air flow signals were alternatively measured from the ventilator. In addition, Hilbert transform was employed to determine the period of lung sounds. The clinical results showed the windowing method could attenuate interference from the heart sounds and enhance the lung sounds. Furthermore, combination with Hilbert transform is an efficient approach to determine the breathing period for long-term monitoring ventilated patients. The study suggests that this system is valid for objective measurement of lung sounds, regardless of patients without or with ventilator support. One of future application of the patients is anticipated to combine real-time monitor and diagnosis for patients with respiratory diseases.

[1] John F, Murray, Jay A, Nadel, “Textbook of respiratory medicine,” Philadelphia: W. B. Saunders, 1994
[2] Leblanc P, Macklin PT, Ross WR, “Breath sounds and distribution of pulmonary ventilation,” Am. Rev. Respir. vol. 102, pp.10-16, 1970
[3] L. Vannuccini, J. E. Earis, P. Helisto, B. M. G. Cheetham, M. Rossi, A. R. A. Sovijarvi, and J. Vanderschoot, “Capturing and preprocessing of respiratory sounds,” Eur. Respir. Vol. 10, no. 77, pp. 616-620, 2000
[4] A. R. A. Sovijarvi, L. P. Malmberg, G. Charbonneau, J. Vanderschoot, F. Dalmasso, C. Sacco, M. Rossi, J. E. Earis, “Characteristics of breath sounds and adventitious respiratory sounds, ” Eur. Respir. Vol. 10, no. 77, pp. 591-596, 2000
[5] F. Dalmay, M. T. Antonini, P. Marquet, R. Menier, “Acoustic properties of the normal chest,” Eur, Respir, Vol. 8, pp.1761-1769, 1995.
[6] Vander, Sherman, and Luciano’s, Human Physiology 9/e, McGrawHill, 2005.
[7] I. Sen and Y. P. Kahya, “A multichannel analog processing circuit for respiratory sound acquisition applications,” Proceedings of the Annual International Conference of the IEEE, Vol.4, pp.3192-3195, 2003
[8] Huseyin Polat and Inan Guler, “A simple computer-based measurement and analysis system of pulmonary auscultation sounds,” Journal of Medical System, Vol. 28, No. 6, 2004
[9] V. Iyer, P. Ramamorthy, H. Fan, and Y. Ploysongsang, “Reduction of heart sounds from respiratory sounds by adaptive filtering,” IEEE Trans. Biomed. Eng., Vol. 33, no. 12, pp. 1141-1148, 1986
[10] L. Hadjileontiadis and S. Panas, “Adaptive reduction of heart sounds from lung sounds using forth-order statistics,” IEEE Trans. Biomed. Eng., Vol. 44, no. 7, pp. 642-648, 1997
[11] I. Hossain and Z. Moussavi, “An overview of heart-noise reduction of lung sound using wavelet transform based filter, ” in Proc. IEEE Eng. Med. Biol. Soc., pp. 458-461, 2003
[12] Z. Moussavi, D.Flores, and G. Thomas, “Heart sound cancellation based on multi-scale products and linear prediction,” in IEEE EMBS, Sept. pp. 3840-3843, 2004
[13] M. T. Pourazad, Z. Moussavi, F. Farahmand, R. K. Ward, “Heart sounds separation from lung sounds using independent component analysis,” Proc. 27th Ann. Int. Conf. IEEE EMBS, Sep. 2005
[14] A. Hyvarinen, “Survey on independent component analysis,” Neural Computing Surveys, Vol. 2, pp. 94-128, 1999
[15] A. Hyvarinen, “Fast and robust fixed-point algorithms for independent component analysis,” IEEE Transactions on Neural Networks, Vol. 10, pp. 626-634, 1999.
[16] A. Hyvärinen, “What is independent component analysis ?”
http://www.cis.hut.fi/projects/ica/icademo/
64