肺癌是目前全球十大死因之首，死亡率極高的原因，大多數來自於晚期發現而延誤了治療，由於初期的癌腫體積太小，現在一般醫療院所的診斷方法難以發現且檢查手續繁複，為了能夠達到快速篩選的效果，此研究利用仿生電子鼻技術，藉由肺癌患者所呼出的氣體，找尋肺癌氣體的特徵，來進而幫助醫生及早判斷肺癌病患的可能性，降低晚期發現的風險。
本研究著重於系統架構，提出了固定流速，金屬腔體的電子鼻系統，並在系統前端加入了乾燥劑管，讓濕度和流速壓力的影響降至最低。利用半導體氧化感測器得到病患所呼出的氣體反應後，首先藉由訊號前處理，處理基線問題，得到資料特徵點。為了減少後續演算法運算複雜，使用主成分分析法和線性判別分析降低維度，降維後也方便觀察資料分布情形。辨識方法使用最近K個鄰居法和支持向量機，搭配留一交叉驗證，兩者互相比較，再以接收者操作曲線進行分類器評估。初始辨識結果約可達到70%，在辨識過程中發現感測器對於時間的飄移日趨嚴重，故使用線性迴歸預測證明感測器的飄移現象如本研究假設所示，再利用直接校正法進行校正，校正後會達到更好的分群效果，最終辨識結果可達到80%~85%。

Lung cancer was currently the top ten causes of death in the world. The reason for the extremely high mortality rate is mostly due to late detection and delayed treatment. Because the initial cancer volume was too small, the diagnostic methods of general medical institutions are difficult to find and the inspection procedures are complicated. In order to achieve rapid screening, we used electronic nose technology to find the characteristics of lung cancer gas through the gas exhaled by lung cancer patients. In order to help doctors to determine the possibility of lung cancer patients early and reduce the risk of late detection.
The research focuses on the system architecture. We proposed constant flow and metallic chamber system. A desiccant tube was added to the front of the system to minimize the effects of humidity and flow pressure. First, we got feature point by signal pre-processing. In order to reduce the complexity of the subsequent algorithm, the principal component analysis (PCA) and linear discriminant analysis (LDA) are used to reduce the dimension, and it is convenient to observe the data distribution. The identification method uses the nearest K neighbor methods and support vector machines, with a leave-one-out cross-validation. Then, we used receiver operating curve for classifier evaluation.

The initial recognition rate was about 65%. During the identification process, it was found that the sensors drift for time. We used linear regression to verify sensors drift and used direct standardization to calibrate sensors drift. After calibration, we can achieve better result. Finally, the recognition rate was about 80%~85%.

目錄
摘要 i
Abstract ii
致謝 iv
目錄 v
第一章 緒論 1
1.1 研究背景與動機 1
1.2 氣體感測系統基本架構 3
第2章 文獻回顧 6
2.1氣體檢測相關文獻討論 6
2.2肺部疾病檢測相關文獻討論 7
第3章 實驗系統架構 9
3.1 訊號擷取系統 10
3.1.1 氣體感測器 10
3.1.2 感測器介面電路 13
3.1.3 感測器腔體 15
3.2 實驗方法與流程 17
3.2.1 氣袋採樣 17
3.2.2 實驗流程 18
第4章 訊號處理與資料分析 20
4.1 訊號前處理 21
4.1.1基線處理 21
4.2 特徵擷取 23
4.2.1 主成分分析 (Principal Components Analysis, PCA) 24
4.2.2 線性判別分析 (Linear Discriminant Analysis, LDA) 25
4.3 分類演算法 27
4.3.1 K個最近鄰居判別 (K-Nearest Neighbors, KNN) 27
4.3.2 支持向量機 (Support Vector Machine, SVM) 28
4.4 交叉驗證方法 31
4.5 線性迴歸預測與直接校正法 32
4.5.1 線性迴歸 32
4.5.1 直接校正法 33
4.6 接收者操作曲線 36
第5章 實驗結果與討論 38
5.1原始資料實驗結果 39
5.1.1資料分佈 39
5.1.2 辨識率結果 43
5.1.3 感測器誤差 47
5.2 校正後資料實驗結果 51
5.2.1 資料分佈 51
5.2.2 線性迴歸預測 54
5.2.3 辨識率與校正結果 59
5.2.4 分類器評估 62
5.3 校正與辨識分析 65
第6章 結論未來發展 66
6.1 結論 66
6.2 未來發展 67
參考文獻 69

參考文獻
[1] 衛生福利部 https://www.mohw.gov.tw/cp-16-33598-1.html
[2]彰化基督教醫院http://www2.cch.org.tw/lungcancer/LC_TX_SUM.htm
[3] Rachel Herz. ”氣味之謎：主宰人類現在與未來生存的神祕感官”
[4]Tim C. Pearce PhD, Prof. Susan S. Schiffman Prof. H. Troy Nagle and Prof. Julian
W. Gardner,” Handbook of Machine Olfaction: Electronic Nose Technology”
[5] Alireza Sanaeifara, Hassan ZakiDizajib, Abdolabbas Jafaria and Miguel de la
Guardia, “Early detection of contamination and defect in foodstuffs by
electronicnose: A review ”
[6] Romero-Flores, Adrian, et al. "Evaluation of an electronic nose for odorant and
process monitoring of alkaline-stabilized biosolids production." Chemosphere 186
(2017): 151-159.
[7] http://chs.ctust.edu.tw/ezfiles/20/1020/attach/20/pta_6325_9538958_39308.pdf
[8] Li, Mingxiao, et al. "Breath carbonyl compounds as biomarkers of lung cancer."
Lung Cancer 90.1 (2015): 92-97.
[9] Konvalina, Gady, and Hossam Haick. "Sensors for breath testing: from
nanomaterials to comprehensive disease detection." Accounts of chemical
research 47.1 (2013): 66-76.

[10] Hockstein, Neil G., et al. "Diagnosis of pneumonia with an electronic nose:
correlation of vapor signature with chest computed tomography scan findings."
The Laryngoscope 114.10 (2004): 1701-1705.
[11] MacNee, William, et al. "Evaluation of exhaled breath condensate pH as a
biomarker for COPD." Respiratory medicine 105.7 (2011): 1037-1045.
[12] Machado, Roberto F., et al. "Detection of lung cancer by sensor array analyses of
exhaled breath." American journal of respiratory and critical care medicine 171.11 (2005): 1286-1291.
[13] Dragonieri, Silvano, et al. "An electronic nose in the discrimination of patients
with non-small cell lung cancer and COPD." Lung cancer 64.2 (2009): 166-170.
[14] Adiguzel, Yekbun, and Haluk Kulah. "Breath sensors for lung cancer diagnosis."
Biosensors and Bioelectronics 65 (2015): 121-138.
[15] Adiguzel, Yekbun, and Haluk Kulah. "Breath sensors for lung cancer diagnosis."
Biosensors and Bioelectronics 65 (2015): 121-138.
[16] 李宥瑾, ” 利用機器學習方法分析電子氣體感測資料以鑑別慢性肺阻塞與氣喘患者” 國立清華大學電機工程研究所, 碩士論文,2015
[17] Montuschi, Paolo, et al. "The electronic nose in respiratory medicine."
Respiration 85.1 (2013): 72-84.

[18] Gregis, Geoffrey, et al. "Detection and quantification of lung cancer biomarkers
by a micro-analytical device using a single metal oxide-based gas sensor."
Sensors and Actuators B: Chemical 255 (2018): 391-400.
[19] FIAGRO 公司 http://www.figarosensor.com/
[20] SKC空氣採樣袋 http://www.kohan.com.tw/skc/air_sample_bags.html#3
[21] The MatlabWorks, INC. https://www.mathworks.com/products/matlab.html
[22] Marco, Santiago, and Agustín Gutierrez-Galvez. "Signal and data processing for
machine olfaction and chemical sensing: A review." IEEE Sensors Journal 12.11
(2012): 3189-3214.
[23] Hines, E. L., E. Llobet, and J. W. Gardner. "Electronic noses: a review of signal
processing techniques." IEE Proceedings-Circuits, Devices and Systems 146.6
(1999): 297-310
[24] Nasrabadi, Nasser M. "Pattern recognition and machine learning." Journal of
electronic imaging 16.4 (2007): 049901.
[25] Wold, Svante, Kim Esbensen, and Paul Geladi. "Principal component analysis." Chemometrics and intelligent laboratory systems 2.1-3 (1987): 37-52.
[26] Izenman, Alan Julian. "Linear discriminant analysis." Modern multivariate statistical techniques. Springer, New York, NY, 2013. 237-280.

[27] Soucy, Pascal, and Guy W. Mineau. "A simple KNN algorithm for text
categorization." Data Mining, 2001. ICDM 2001, Proceedings IEEE
International Conference on. IEEE, 2001.
[28] Smola, Alex J., and Bernhard Schölkopf. "A tutorial on support vector
regression." Statistics and computing 14.3 (2004): 199-222.
[29] Kohavi, Ron. "A study of cross-validation and bootstrap for accuracy estimation
and model selection." Ijcai. Vol. 14. No. 2. 1995.
[30] Seber, George AF, and Alan J. Lee. Linear regression analysis. Vol. 329. John
Wiley & Sons, 2012.
[31] Tomic, Oliver, Heiko Ulmer, and John-Erik Haugen. "Standardization methods
for handling instrument related signal shift in gas-sensor array measurement
data." Analytica Chimica Acta 472.1-2 (2002): 99-111.
[32] Hanley, James A., and Barbara J. McNeil. "The meaning and use of the area
under a receiver operating characteristic (ROC) curve." Radiology 143.1 (1982):
29-36.