加護病房中存在高抗藥性肺炎病菌的威脅，若病人感染肺炎，則需要診斷為何種肺炎菌種才能對症下藥，而目前普遍採取菌種培養的診斷方法，但此方法需要約五到六天才能得知結果，往往緩不濟急，病人就去世了。因此，想將具肺炎菌種辨認功能的電子鼻晶片安裝於人工呼吸器上，隨時偵測病人呼出的氣體中是否含有肺炎菌種氣體成分，以幫助醫生作早期診斷並對症下藥，把握治療的黃金時期。本論文研究過程中使用奈米複合材料陣列感測器，蒐集病人呼出的氣體訊號，使用樣式識別方法來分析資料。此論文的資料分析方面，使用KNN作為分類器，並且使用循序的特徵選擇方法挑出對肺炎辨識具有影響力的感測器以提高辨識率。實驗結果顯示，肺炎偵測使用特徵選擇方法，正確率從73%稍微上升至75%；而肺炎菌種辨認正確率則從66%明顯提高至73%。
論文中也提出基於構成成分作決策的混合氣體辨識方法(Individual Constituent-Decision Method, ICDM)，此方法不只將混合的涵義包含在內，且可以針對各構成成分的辨識作最佳化，將所有的構成成分決策結果綜合起來，即為混合氣體的辨識結果。由於本研究尚未有肺炎菌種氣體的主要構成成分資料，所以使用市售的混合果汁氣體資料來驗證此方法。此論文將ICDM和一次把混合後的所有組合作分類的方法比較，實驗結果發現ICDM有比較高的辨識率，並且各構成成分的辨識模型(最佳特徵子集合、分類器參數)皆有所不同，表示ICDM能針對各構成成分的辨識作最佳化。

There is a serious threat of high-resistance pneumonia bacteria in the Intensive Care Unit. If a doctor finds out patients infected with pneumonia, he needs to determine which types of bacteria causes the trouble in order to prescribe the right antibiotic medicine. Up to now bacteria culture is the common way of diagnosis, but it takes five to six days to get the results. This is usually slow for emergency and some patients do not survive during the wait. Therefore, the idea of installing an electronic nose with pneumonia bacteria recognition function on artificial respiration is born. The electronic nose aims to detect whether the constituents of the gas exhaled by the patient imply bacterial infection in the lung. The monitoring can be done continually, helping physicians to perform early diagnosis and prescribe the right medicine, grasping the prime-time on saving the patient’s life.
Nano composite-array sensors are used here to get the signal from exhaled gas by the patients. Pattern recognition approaches were adopted to analyze the data; in this thesis, we choose the K nearest neighbor method (KNN) as our classifier and use sequential feature selection to obtain features that are most effective in discriminating between different types of pneumonia bacteria. The results show that the recognition rate of pneumonia detection increased slightly from 73% to 75% and the recognition rate of pneumonia bacteria recognition improved from 66% to 73%, thanks to the sequential feature selection.
This thesis also proposes a novel mixture gas recognition method, which we call the Individual Constituent-Decision Method (ICDM). The method utilizes the physical meaning of mixture, and it can be optimized separately to detect each constituent of interest. Results from all constituent-decision makers can be combined so as to produce a final result of mixture gas recognition. To validate ICDM in this thesis, because the constituting gases are unknown in the gas exhaled from pneumonia patients, we use fruit juice mixtures instead, to emulate the scenario of mixture-gas sensing. We compare ICDM with the traditional method that aims to classify all mixture combinations at one shot. Results show that ICDM has a better performance because it can find different recognition models (the best feature subset and parameters of the classifiers) for each individual constituent. This validates the idea that ICDM should be able to optimize on each individual constituent-decision.
 

摘要 i
Abstract ii
誌謝 iv
第一章 緒論 1
1.1生物嗅覺機制 1
1.2電子鼻介紹 2
1.3 肺炎簡介以及文獻回顧 2
1.4研究動機與目的 4
第二章 奈米複合材料陣列感測器介紹及資料取得與分析 6
2.1奈米複合材料陣列感測器介紹 6
2.2臨床資料取得方式及使用資料介紹 7
2.3肺炎氣體資料於GC-MS分析結果 8
第三章 資料辨識演算法 10
3.1資料前處理 11
3.1.1 基線操縱 11
3.1.2 暫態壓縮 13
3.1.3 正規化 14
3.2 特徵選擇演算法 15
3.2.1 維度的詛咒以及維度縮減 15
3.2.2 特徵選擇：循序向後的特徵選擇演算法(SBS) 16
3.2.3 特徵選擇：循序浮動式向後的特徵選擇演算法(SFBS) 20
3.3分類器 21
3.3.1 K個最鄰近點分類器 21
3.3.2 在鄰近點使用權重的K個最鄰近點分類器 23
3.4 模型訓練 24
3.5 效果評估以及驗證 25
3.5.1 正確率 26
3.5.2 F1測量 26
3.5.3 交叉驗證 27
3.6 混合氣體辨識方法 28
3.6.1多類別辨識方法(Multi-Class Classification) 28
3.6.2基於構成成分作決策的混合氣體辨識方法(Individual Constituent-Decision Method, ICDM) 29
第四章 結果分析與討論 32
4.1 肺炎菌種氣體分析 32
4.1.1 肺炎偵測 33
4.1.2 肺炎菌種辨識 39
4.2 混合氣體辨識方法分析 44
4.2.1 多類別的辨識方法：一次對混合後的所有組合作分類 45
4.2.2 基於構成成分作決策的辨識方法(ICDM) 46
第五章 結論與未來工作 51

[1] 林怡文, “一「嗅」萬千的巧妙,” 科學人雜誌, 2012. [Online]. Available: http://sa.ylib.com/MagCont.aspx?Unit=easylearn&id=1900.
[2] K. Persaud and G. Dodd, “Analysis of discrimination mechanisms in the mammalian olfactory system using a model nose,” Nature, vol. 299, no. 5881, pp. 352–355, Sep. 1982.
[3] J. W. Gardner and P. N. Bartlett, “A brief history of electronic noses,” Sensors and Actuators B: Chemical, vol. 18, no. 1–3, pp. 210–211, Mar. 1994.
[4] P. C. Jain and R. Kushwaha, “Wireless gas sensor network for detection and monitoring of harmful gases in utility areas and industries,” in 2012 Sixth International Conference on Sensing Technology (ICST), 2012, pp. 642–646.
[5] Y. Yin, “Discrimination between Chinese Jing Wine and Counterfeit Using Different Signal Features of an Electronic Nose,” Journal of Sensor Technology, vol. 02, no. 03, pp. 109–115, 2012.
[6] J. W. Gardner, H. V. Shurmer, and T. T. Tan, “Application of an electronic nose to the discrimination of coffees,” Sensors and Actuators B: Chemical, vol. 6, no. 1–3, pp. 71–75, Jan. 1992.
[7] 周亦斌 and 王俊, “基于电子鼻的番茄成熟度及贮藏时间评价的研究,” 农业工程学报, vol. 21, no. 4, pp. 113–117, 2005.
[8] C. Di Natale, A. Macagnano, E. Martinelli, R. Paolesse, G. D’Arcangelo, C. Roscioni, A. Finazzi-Agrò, and A. D’Amico, “Lung cancer identification by the analysis of breath by means of an array of non-selective gas sensors,” Biosensors and Bioelectronics, vol. 18, no. 10, pp. 1209–1218, Sep. 2003.
[9] J. W. Gardner, M. Craven, C. Dow, and E. L. Hines, “The prediction of bacteria type and culture growth phase by an electronic nose with a multi-layer perceptron network,” Measurement Science and Technology, vol. 9, no. 1, pp. 120–127, Jan. 1998.
[10] 行政院衛生署, “民國100年國人前十大死因.” [Online]. Available: http://www.doh.gov.tw/cht2006/index_populace.aspx.
[11] D. K. Iakovidis, S. Tsevas, M. A. Savelonas, and G. Papamichalis, “Image analysis framework for infection monitoring.,” IEEE transactions on bio-medical engineering, vol. 59, no. 4, pp. 1135–44, May 2012.
[12] 鄭桂忠, 施崇鴻, 王立群, 陳新, 劉奕汶, 徐爵民, 楊家銘, and 饒達仁, “以電子鼻系統晶片早期預測及同步診斷使用人工呼吸器病患的肺炎菌種,” 國科會計畫 NSC 101-2220-E-007 -0062012.
[13] C. W. Hanson and E. R. Thaler, “Electronic Nose Prediction of a Clinical Pneumonia Score: Biosensors and Microbes,” Anesthesiology, vol. 102, no. 1, pp. 63–68, Jan. 2005.
[14] N. G. Hockstein, E. R. Thaler, D. Torigian, W. T. Miller, O. Deffenderfer, and C. W. Hanson, “Diagnosis of pneumonia with an electronic nose: correlation of vapor signature with chest computed tomography scan findings.,” The Laryngoscope, vol. 114, no. 10, pp. 1701–5, Oct. 2004.
[15] Intelligent Optical Systems, “Cyranose Overview.” [Online]. Available: http://www.intopsys.com/products/eNose Sensors Overview.pdf.
[16] H. Bai and G. Shi, “Gas Sensors Based on Conducting Polymers,” Sensors, vol. 7, no. 3, pp. 267–307, Mar. 2007.
[17] “GCMS - How Does It Work?,” Oregon State University Environmental Health Sciences Center. [Online]. Available: http://www.unsolvedmysteries.oregonstate.edu/MS_05.
[18] H. H. Maurer, “Systematic toxicological analysis of drugs and their metabolites by gas chromatography—mass spectrometry,” Journal of Chromatography B: Biomedical Sciences and Applications, vol. 580, no. 1–2, pp. 3–41, Sep. 1992.
[19] M. Tsivou, N. Kioukia-Fougia, E. Lyris, Y. Aggelis, A. Fragkaki, X. Kiousi, P. Simitsek, H. Dimopoulou, I.-P. Leontiou, M. Stamou, M.-H. Spyridaki, and C. Georgakopoulos, “An overview of the doping control analysis during the Olympic Games of 2004 in Athens, Greece,” Analytica Chimica Acta, vol. 555, no. 1, pp. 1–13, Jan. 2006.
[20] J. R. Wallace, “GC/MS data from fire debris samples: interpretation and applications,” Journal of forensic sciences, vol. 44, no. 5, pp. 996–1012, 1999.
[21] A. Stambouli, A. El Bouri, T. Bouayoun, and M. A. Bellimam, “Headspace-GC/MS detection of TATP traces in post-explosion debris.,” Forensic science international, vol. 146 Suppl, pp. S191–4, Dec. 2004.
[22] R. Gutierrez-Osuna and H. T. Nagle, “A method for evaluating data-preprocessing techniques for odour classification with an array of gas sensors.,” IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics : a publication of the IEEE Systems, Man, and Cybernetics Society, vol. 29, no. 5, pp. 626–32, Jan. 1999.
[23] T. C. Pearce, Handbook of Machine Olfaction. Weinheim, FRG: Wiley-VCH Verlag GmbH & Co. KGaA, 2002.
[24] R. Gutierrez-Osuna, “Pattern analysis for machine olfaction: a review,” IEEE Sensors Journal, vol. 2, no. 3, pp. 189–202, Jun. 2002.
[25] E. Llobet, J. Brezmes, X. Vilanova, J. E. Sueiras, and X. Correig, “Qualitative and quantitative analysis of volatile organic compounds using transient and steady-state responses of a thick-film tin oxide gas sensor array,” Sensors and Actuators B: Chemical, vol. 41, no. 1–3, pp. 13–21, Jun. 1997.
[26] T. Eklöv, P. Mårtensson, and I. Lundström, “Selection of variables for interpreting multivariate gas sensor data,” Analytica Chimica Acta, vol. 381, no. 2–3, pp. 221–232, Feb. 1999.
[27] G. Hughes, “On the mean accuracy of statistical pattern recognizers,” IEEE Transactions on Information Theory, vol. 14, no. 1, pp. 55–63, Jan. 1968.
[28] S. Wold, K. Esbensen, and P. Geladi, “Principal component analysis,” Chemometrics and Intelligent Laboratory Systems, vol. 2, no. 1–3, pp. 37–52, Aug. 1987.
[29] R. A. Fisher, “The use of multiple measurements in taxonomic problems,” Annals of Eugenics, vol. 7, no. 2, pp. 179–188, Sep. 1936.
[30] P. A. Devijver and J. Kittler, Pattern recognition: A statistical approach. Prentice/Hall International Englewood Cliffs, NJ, 1982.
[31] P. Pudil, J. Novovičová, and J. Kittler, “Floating search methods in feature selection,” Pattern Recognition Letters, vol. 15, no. 11, pp. 1119–1125, Nov. 1994.
[32] M. Dash and H. Liu, “Feature selection for classification,” Intelligent Data Analysis, vol. 1, no. 1–4, pp. 131–156, 1997.
[33] C. C. Reyes-Aldasoro and A. Bhalerao, “The Bhattacharyya space for feature selection and its application to texture segmentation,” Pattern Recognition, vol. 39, no. 5, pp. 812–826, May 2006.
[34] G.-O. Ricardo, “sequential feature selection,” Computer Science and Engineering, Texas A&M University. [Online]. Available: http://research.cs.tamu.edu/prism/lectures/pr/pr_l11.pdf.
[35] Burton DeWilde, “Classification of Hand-written Digits,” 2012. [Online]. Available: http://datasciencerules.blogspot.tw/2012/10/classification-of-hand- written-digit s-3.html.
[36] R. O. Duda, P. E. Hart, and D. G. Stork, Pattern classification, 2nd ed. Wiley, 2001.
[37] S. Tan, “Neighbor-weighted K-nearest neighbor for unbalanced text corpus,” Expert Systems with Applications, vol. 28, no. 4, pp. 667–671, May 2005.
[38] G. H. John, R. Kohavi, and K. Pfleger, “Irrelevant features and the subset selection problem,” in Proceedings of the eleventh international conference on machine learning, 1994, vol. 129, pp. 121–129.
[39] S. Theodoridis and K. Koutroumbas, Pattern Recognition, Fourth Edition, 4th ed. Academic Press, 2008, pp. 280–282.
[40] 野原三十郎, “Precision and Recall-Informaiton Retreival-2,” 2009. [Online]. Available: http://ims.tw/archives/553.
[41] D. M. Magerman, “Statistical decision-tree models for parsing,” in Proceedings of the 33rd annual meeting on Association for Computational Linguistics -, 1995, pp. 276–283.
[42] G. H. Kim, Y. W. Kim, S. J. Lee, and G. J. Jeon, “Multi-class system based on SVM for real-time gas mixture classification,” in SICE Annual Conference 2010, Proceedings of, 2010, pp. 1764–1767.
[43] 許柏安, 快速混合氣體辨識方法之研究, 國立清華大學資訊工程研究所, 碩士論文, 2012.