腦波圖是一個對於研究或是臨床診斷時，常見且具說服力的檢測工具之一。特別是當時間分辨率需要毫秒以內的檢查，更能展現出它的價值。由於腦波圖中蘊含了身體內部豐富的資訊，且具有低成本和非侵入式檢測的優點，所以這項工具常用來檢測如癲癇症或其他腦部相關之疾病。然而，由於腦波圖的電極是直接黏貼在頭皮上，所以很容易受到環境因素的影響。另外，腦波圖包含了 20 條以上的頻道數目，而且每一個記錄都是長達 30 分鐘以上的資料量。而對於如此龐大的數據，能夠反映出身體狀況的資訊卻蘊含在每分每秒之中，所以醫生在判讀腦波圖的時候，需要花費很多時間與精神，不過也容易造成不必要的錯誤與遺漏。因此，我們將目標鎖定在醫師判讀腦波圖的時候，會觀察的條件之一。那就是有關大腦對稱性的問題。由於我們人體正常情況下，都是處於平衡且對稱的狀態。如果我們腦波有不對稱的情況發生時，那就表示我們身體某處機能很可能出了問題 ，所以我們想自動找出腦波不對稱的區間。因此，在此論文中，我們提出兩種方法來判斷出腦波圖中不對稱的區間。首先，在我們的第一個方法中，我們基於左右腦之間振幅的差異度比率當作依據。而振幅差異度的計算，我們是採用左、右腦波型下之面積總合並相除來表示。如果差異程度超過設立的門檻值，即判定為不對稱區間。而在我們的運算中，都是專注於大腦前額較敏感的 4 條頻道來做分析。另外，由於腦波圖很容易受到環境因素的影響，時常會有短暫的波峰值產生，而造成假性異常的情況。因此，我們定義每一次的運算區間為非短暫性的 5 秒鐘。這樣能夠濾除短暫性異常的狀況，也能兼具醫師所認定的異常活動皆為持續性的行為之特點。然而對於腦波的不對稱判斷，醫生會考慮的因素除了振幅之外，還包括頻率、波型和其餘特殊情況。若只有考慮振幅一種因素的話，我們將會無法處理所有的情況。因此，在我們的第二個方法中，我們採用支持向量機的學習方式，並考慮了三種因素:振幅差異度、方均根差異度與前額不對稱指數。透過三種特徵的訓練方式，我們可以讓電腦自動化的學習並進行判斷不對稱之區間。最後將我們的結果和醫生觀察的結果做比較，雖然還是無法處理所有情況，但相信倘若隨著訓練資料量的提高，或是找到其他更具代表性的特徵來做訓練，其結果會有更好的成效。

The electroencephalography is a common and valuable tool for research and diagnosis, especially when millisecond-range temporal resolution is required. EEG is usually used to diagnose epilepsy and other diseases of brain. This method has low cost and non-invasive advantages, and contains a lot of information. However, the electrodes are attached on the scalp directly, so the impact of the environment is obvious. Moreover, the EEG recordings contain more than 20 channels in a test and each test records for at least 30 minutes. The EEG can reflect the signs of physical conditions in every minute so the neurologists will spend a lot of time and spirit for the interpretation of the EEG recordings. There will easily be errors and omission in the interpretation. Therefore, we focus on a condition which the neurologists will pay attention to in the interpretation of EEG recordings. That is about the brain symmetry problem because our body is balanced and in the symmetrical state. If we have asymmetric situation in the brain, it means that we have the possibility of the functional abnormalities. Therefore, in this thesis, we propose two methods to determine the interval of asymmetry. The first one is the Amplitude Difference Ratio Method (ADR), and we are based on the ratio of the amplitude between left and right hemisphere. We divide the total area under the waveforms of left and that of right hemisphere as the ratio of asymmetry. If the degree of the difference is over a threshold which we set, this interval will be marked as an asymmetry interval. Basically, our methods focus on the frontal channels of the brain to be analyze, because the front of the channels are more sensitive to the electrode positions. Otherwise, due to the EEG is usually impacted by the environment, there are peak waveforms occurred as the false abnormal cases. Therefore, we define the size of intervals as 5 seconds which is the non-short interval while computing. It can filter out the short period of waveforms and confirm that the neurologists believe that abnormal activities are persistent behavior. However, the physicians does not only consider the amplitude, but also consider the frequency, waveforms, and other specific cases for interpreting the asymmetry. If we only consider the amplitude, we could not handle all of asymmetry cases. Therefore, our second method uses the Support Vector Machine with 3 Features (SVM3F) for learning, and considers three factors: the ratio of amplitude, the ratio of root mean square and the frontal asymmetry index. We can make the computer automatically learning to detect the intervals of asymmetry. Finally, we compare our results with the physician’s observation. Although we cannot handle all cases, we have some suggestions. The larger the number of training data is, or the representative features have, the better accuracy of the results is.

[THESIS VALIDATIONLETTER+i]
[ACKNOWLEDGEMENTS+ii]
[ABSTRACT(CHINESE)+iii]
[ABSTRACT(ENGLISH)+iv]
[LIST OF FIGURES+vii]
[LIST OF TABLES+ix]
[1. Introduction+1]
[1.1 Electroencephalography+2]
[1.2 Motivation+7]
[1.3 Organization of the Thesis+8]
[2. Background+9]
[2.1 Electroencephalogram+9]
[2.1.1 Electrodes Location+11]
[2.1.2 Montages+14]
[2.1.3 Reading EEG+17]
[3. The Proposed Methods+22]
[3.1 The Classification of 23 Channels+22]
[3.2 Filter+26]
[3.3 The Amplitude Difference Ratio Method (ADR)+29]
[3.4 The Support Vector Machine with 3 Features Method (SVM3F)+35]
[3.4.1 Root Mean Square+35]
[3.4.2 Frontal Asymmetry Index+36]
[3.4.3 Support Vector Machine+37]
[3.5 Conditions+37]
[4. Performance+41]
[4.1 Performance Model+41]
[4.1.1 Study Design+43]
[4.1.2 Performance Measures+43]
[4.1.3 Database+44]
[4.2 Experiments Results+45]
[4.2.1 Amplitude Difference Ratio+45]
[4.2.2 Support Vector Machine with 3 Features+51]
[5. Conclusion+54]
[5.1 Summary+54]
[5.2 Future Work+55]
[BIBLIOGRAPHY+56]

[1] “American Clinical Neurophysiology Society,” http://www.acns.org/.
[2] “CHB-MIT Scalp EEG Database,” https://www.physionet.org/pn6/chbmit/.
[3] “EEGLAB Toolbox,” https://sccn.ucsd.edu/eeglab/.
[4] “The european data format,” http://www.edfplus.info/.
[5] “A free, opensource, multiplatform, universal viewer and toolbox intended for, but not limited to, timeseries storage files like eeg, emg, ecg, bioimpedance, etc.,”http://www.teuniz.net/edfbrowser/.
[6] A. A. Abdul-latif, I. Cosic, D. K. Kumar, B. Polus, and C. D. Costa, “Power Changes of EEG Signals Associated with Muscle Fatigue: The Root Mean Square Analysis of EEG Bands,” Proceedings of Intelligent Sensors, Sensor Networks and Information Processing Conference, pp. 531–534, Dec. 2004.
[7] U. R. Acharya, F. Molinari, S. V. Sree, S. Chattopadhyay, K. H. Ng, and J. S. Suri, “Automated Diagnosis of Epileptic EEG Using Entropies,” Biomedical Signal Processing and Control, Vol. 7, No. 4, pp. 401–408, July 2012.
[8] U. R. Acharya, S. V. Sree, A. P. C. Alvin, and J. S. Suri, “Use of Principal Component Analysis for Automatic Classification of Epileptic EEG Activities in Wavelet Framework,” Expert Systems with Applications, Vol. 39, No. 10, pp. 9072–9078, Aug. 2012.
[9] U. R. Acharya, S. V. Sree, G. Swapna, R. J. Martis, and J. S. Suri, “Automated EEG Analysis of Epilepsy: A Review,” Knowledge-Based Systems, Vol. 45, pp. 147–165, June 2013.
[10] H. Adeli, Z. Zhou, and N. Dadmehr, “Analysis of EEG Records in an Epileptic Patient Using Wavelet Transform,” Journal of Neuroscience Methods, Vol. 123, No. 1, pp. 69–87, Feb. 2003.
[11] A. Anwar, “An Entropy-based Feature in Epileptic Seizure Prediction Algorithm,” IOSR-Journal of Computer Engineering, Vol. 17, No. 6, pp. 47–54, Dec. 2015.
[12] C.-C. Chang and C.-J. Lin, “LIBSVM: A Library for Support Vector Machines,” ACM Transactions on Intelligent Systems and Technology, Vol. 2, No. 3, pp. 27:1–27:27, Apr. 2011.
[13] J. A. Coan and J. J. Allen, “Frontal EEG Asymmetry and the Behavioral Activation and Inhibition Systems,” Society for Psychophysiological Research, Vol. 40, No. 1, pp. 106–114, Feb. 2003.
[14] J. Coan and J. J. B. Allen, “The State and Trait Nature of Frontal EEG Asymmetry in Emotion,” The Asymmetrical Brain, p. 565V615, Jan. 2003.
[15] A. Goldberger, L. Amaral, L. Glass, J. Hausdorff, P. Ivanov, R. Mark, J. Mietus, G. Moody, C. Peng, and H. Stanley, “Physiobank, Physiotoolkit, and Physionet: Components of a New Research Resource for Complex Physiologic Signals,” Circulation, Vol. 101, No. 23, pp. 215–220, June 2000.
[16] H. Jasper, “Report of the Committee on Methods of Clinical Examination in Electroencephalography,” Electroencephalography and Clinical Neurophysiology, Vol. 10, No. 2, pp. 370–375, May 1958.
[17] R. P. L. Jr., “Brain Function and Deficits,” http://www.neuroskills.com/brain-injury/brain-function.php.
[18] J. Malmivuo and R. Plonsey, “Bioelectromagnetism: Principles and Applications of Bioelectric and Biomagnetic Fields,” http://www.bem.fi/book/00/tx.htm, 1995.
[19] M. Mohammadi, F. Al-Azab, B. Raahemi, G. Richards, N. Jaworska, D. Smith, S. de la Salle, P. Blier, and V. Knott, “Data Mining EEG Signals in Depression for Their Diagnostic Value,” BMC Medical Informatics and Decision Making, Vol. 15, No. 1, pp. 1–14, Dec. 2015.
[20] I. Mporas, V. Tsirka, E. I. Zacharaki, M. Koutroumanidis, M. Richardson, and V. Megalooikonomou, “Seizure Detection Using EEG and ECG Signals for Computer-based Monitoring, Analysis and Management of Epileptic Patients,” Expert Systems with Applications, Vol. 42, No. 6, pp. 3227–3233, April 2015.
[21] M. Z. Parvez and M. Paul, “Epileptic Seizure Detection by Analyzing EEG Signals Using Different Transformation Techniques,” Neurocomputing, Vol. 145, pp. 190–200, Dec. 2014.
[22] L. Patnaik and O. K. Manyam, “Epileptic EEG Detection Using Neural Networks and Post-classification,” Computer Methods and Programs in Biomedicine, Vol. 91, No. 2, pp. 100–109, Aug. 2008.
[23] E. Pippa, E. I. Zacharaki, I. Mporas, V. Tsirka, M. P. Richardson, M. Koutroumanidis, and V. Megalooikonomou, “Improving Classfication of Epileptic and Non-epileptic EEG Events by Feature Selection,” Neurocomputing, Vol. 171, pp. 576–585, Jan. 2016.
[24] Shoeb and A. Hossam, “Application of Machine Learning to Epileptic Seizure Onset Detection and Treatment,” Massachusetts Institute of Technology, 2009.
[25] E. Smith, “Introduction to EEG,” http://www.ebme.co.uk/articles/clinical-engineering/56-introduction-to-eeg?showall=1.
[26] A. Subasi and M. I. Gursoy, “EEG Signal Classification Using PCA, ICA, LDA and Support Vector Machines,” Expert Systems with Applications, Vol. 37, No. 12, pp. 8659–8666, Dec. 2010.
[27] M. M. S. Sumit Kumar Srivastava, Sharique Ahmed, “Analysis of Brain Signal for the Detection of Epileptic Seizure,” International Journal of Application or Innovation in Engineering and Management, Vol. 5, No. 5, pp. 283–287, May 2016.
[28] A. Tzallas, M. Tsipouras, and D. Fotiadis, “Epileptic Seizure Detection in EEGs Using Time-Frequency Analysis,” IEEE Trans. on Information Technology Biomed, Vol. 13, No. 5, pp. 703–710, Sept. 2009.
[29] P. Valenti, E. Cazamajou, M. Scarpettini, A. Aizemberg, W. Silva, and S. Kochen, “Automatic Detection of Interictal Spikes Using Data Mining Models,” Journal of Neuroscience Methods, Vol. 150, No. 1, pp. 105–110, Jan. 2006.