本論文研究目的，是藉由現今語音訊號的分類與辨識技術的成長，設計與實現肺音量測之人機介面系統此系統先依據特徵值所定義的特性分類出正常肺音與異常肺音，進而辨識出肺部之病症，做為醫師診斷之參考，也提供臨床應用和教學交流之用。在本論文中，對於特徵值擷取、硬體架構、演算法推導以及類神經網路都有完整介紹。
在本研究中針對肺部所發出之哮鳴異常聲音做偵測及判別。哮鳴音是一種偶發性且呈現連續高音調的肺部聲音，當呼吸道收縮產生氣流時，氣流經狹窄氣管發生振動而產生聲音，此聲音中夾雜著咻咻的聲音特徵，所以經常被拿來當作某些肺部疾病的重要特徵之一。
所設計的肺音感測人機介面系統，共包含硬體與軟體兩大部分，硬體架構部份，包含壓電麥克風及資料擷取卡，用來擷取肺部聲音訊號並將類比訊號轉換成數位訊號至電腦端做處理。軟體架構部份，使用MATLAB及LabVIEW程式，先利用MATLAB模擬類神經整體架構之可行性及內部參數，再利用LabVIEW設計訊號前處理及特徵值轉換架構，並進一步整合建構成人機介面用來分析紀錄與顯示判斷其哮鳴音的特徵量測結果。
最後，經由訓練後之類神經網路架構分類，對於肺音訊號是否具有哮鳴異常肺音之判別，其正確率可達92%，並可由所設計之人機介面顯示其肺音波形、特徵值及頻譜分析圖，可供醫生作為診斷肺部疾病病患之輔助用。

The purpose of this paper is presented the way of the speech signal classification and the development of identify technology to establish the Human-Machine interaction system of pulmonary sounds recording process. First, we have to classify the normal pulmonary and the abnormal pulmonary sounds, we can recognize the pulmonary disease. All these researches can help doctors to diagnose and also to provide clinical application and education. This paper will introduce the retrieve of eigenvalue, hardware, derivation of algorithm and the artificial neural network completely.
This paper, is aimed at the research of wheezes. It is a characteristic of the pulmonary disease, like asthma. Wheezes are a kind of pulmonary sounds that is adventitious and continuous, which is probably cause of the vibration on narrow airways when airflow passing through, which is mixed with the sound heards like “shoo”, so we often treat the wheeze as one of the most important characteristic of the pulmonary disease.
The Human-Machine lung sound sensing system which is designed by this research, including two major parts of hardware and software altogether. The hardware architecture including the piezoelectric microphone, materials pick and fetch card, to record lung sounds signal and analogy signal and change them to digital signal for computer to process. The software architecture including the Matlab and LabVIEW programs. This research is used MATLAB to simulate the artificial neural network to recognize the characteristic and learn how to determine the characteristic of the wheeze. And using the LabVIEW to establish the human-machine to analyze the record, and display the result of lung sound recording peocess.
By the research results of this paper, the neural network structure classification via training, the correct rate is up to 92%, and show its lungs sound wave, characteristic value and spectral analysis chart by the human-machine interface design.

[1] http://www.goh.gon.tw/cht2006/index_populace.aspx, 行政院衛生署統計資料, 2008.
[2] Jen-Chien Chien, Ming-Chuan Huang, Yue-Der Lin and Fok-Ching Chong, “A Study of Heart Sound and Lung Sound Separation by Independent Component Analysis Technique,” IEEE International Conference on EMBS, pp. 5708-5711, 2006.
[3] Ipek Sen, Yasemin P. Kahya, “A Multi-Channel Analog Processing Circuit for Respiratory Sound Acquisition Applications,” IEEE International Conference on EMBS, pp. 3192-3195, 2003.
[4] Leontios J. Hadjileontiadis, Stavros M. Panas, “Autoregressive Modeling of Lung Sounds Using Higher-Order Statistics: Estimation of Source and Transmission,” IEEE International Conference on EMBS, pp. 4-8, 1997.
[5] Stavrakoudis, D., Mastorocostas, P. and Theocharis, J., “A Pipelined Recurrent Fuzzy Neural Filter for the Separation of Lung Sounds,” IEEE International Conference on Fuzzy Systems Conference, pp. 1-6, 2007.
[6] P. A. Mastorocostas, J. B. Theocharis, “A stable learning algorithm for block-diagonal recurrent neural network: application to the analysis of lung sounds,” IEEE Transactions on System, pp. 242-254, 2006.
[7] John F. Murray, Jay A. Nadel, “Textbook of respiratory medicine,” Philadelphia: W. B. Saunders, 1994.
[8] 蕭幸宜、蔡秀鸞等合著，「呼吸系統 身體檢查與評估」，台中市：華格那，2005。
[9] 顧潔修, 「理學檢查與健康評估」，藝軒出版社，2003.
[10] A. Kandaswamy, C. Sathish Kumar, Rm. Pl. Ramanathan, S. Jayaraman and N. Malmurugan, “Neural classification of lung sounds using wavelet coefficients,” Computer in Biology and Medicine 34, 523-537, 2004.
[11] Simon Haykin, “Neural Networks: A Comprehensive Foundation, 2nd
Edition,” Prentic Hall International, Inc. 1999.
[12] H. Pasterkamp, S. Kraman and G. Wodicka, “Respiratory Sound-Advance Beyond the Stechoscope,” Am. J. Respir. Crit.. Care. Med., vol. 156, pp. 974-987, 1997.
[13] 蕭子健、周森益、鄭博修、林珮瑜、黃欽章編著，「LabVIEW分析篇」，高立圖書有限公司，2003。
[14] Satish Kumar, “Neural Networks - A Classroom Approach,” McGraw-Hill Inc. 2005.
[15] Hagan, Demuth and Beale, “Neural Network Design”, MHB, Inc. 2004.
[16] 羅華強編著，「類神經網路-Matlab的應用」，高立圖書有限公司，2005。
[17] Satish Kumar, “Neural Networks - A Classroom Approach,” McGraw-Hill Inc. 2005.
[18] Simon Haykin, “Neural Networks: A Comprehensive Foundation,” Prentice Hall International, Inc. 1999.
[19] Hagan Demuth Beale原著，汪惠健審校，「類神經網路設計Neural Network Design」，新加玻商湯姆生亞洲私人有限公司，2004。
[20] 顧力栩、沈晉惠譯，「人工智慧:智慧型系統導論」(第二版)，全華圖書股份有限公司，2007。
[21] 張文恭、江昭皚譯，「運算放大器原理與應用」，格致圖書公司，1991。
[22] Kevin S. Woods, Kevin W. Bowyer, “Generating ROC Curves for Artificial Neural Network,” IEEE Transactions on Medical Imaging, pp. 329-337, 1997.
[22] “A Basic Introduction to Filters –Active, Passive and Switched-Capacitor,” Application Note AN779, National Semiconductor Corporation, 1991.
[23] http://www.biopac.com/Research.asp?Pid=3689&lower=1, BIOPAC Systems, Inc. 2008.
[24] http://zh.wikipedia.org/wiki/%E9%BA%A6%E5%85%8B%E9%A3%8E, 維基百科, 2008.
[25] http://sine.ni.com/nips/cds/view/p/lang/zhs/nid/14903, National Instruments PXI-4472B, 2008.