臺灣博碩士論文加值系統

摘 要 i
ABSTRACT ii
誌 謝 iii
目 錄 iv
表目錄 vii
圖目錄 ix
第一章　緒論 1
1.1 前言與研究動機 1
1.2 文獻回顧 3
1.3 研究方法與目的 5
1.4 論文架構 6
第二章　數據收集 7
2.1 參與研究患者 7
2.2 實驗數據 8
2.2.1 神經學症狀評估NIHSS 10
2.2.2 靜脈血栓溶解劑治療t-PA 10
2.2.3 意識狀況分數GCS score 11
2.2.4 電腦斷層缺血分數ASPECTS 11
2.2.5 手術結果分級mTICI 12
2.2.6 神經功能評估等級mRS 12
2.3 圖像數據 13
2.3.1 腦部電腦斷層影像CT圖 13
2.3.2 施打顯影劑後之腦部電腦斷層影像CTA圖 14
第三章　研究方法 16
3.1 機器學習 16
3.1.1 K近鄰演算法 16
3.1.2 決策樹 17
3.1.3 隨機森林 18
3.1.4 支持向量機SVM 19
3.1.5 RUSBoost 21
3.1.6 CatBoost 22
3.2 交叉驗證 24
3.3 深度學習CNN 25
3.3.1 ResNet50 26
3.3.2 Inception V3 27
第四章　數據處理 29
4.1 數據不平衡 29
4.1.1 Bagging 29
4.1.2 過採樣 29
4.1.3 欠採樣 31
4.2 電腦斷層影像處理 32
4.3 正規化 33
第五章　實驗結果與討論 34
5.1 實驗架構與資料挑選 34
5.2 實驗流程 35
5.3 採納82位病患 36
5.3.1 純臨床資料預測結果 36
5.3.2 臨床資料加入CT、CTA資訊預測結果 37
5.3.2.1 加入CNN預測結果 38
5.3.2.2 擷取CNN預測結果的前一神經層 40
5.3.3 醫師挑選15項參數預測結果 45
5.3.3.1 加入CNN預測結果 46
5.3.3.2 擷取CNN預測結果的前一神經層 49
5.4 排除手術較差患者 51
5.4.1 純臨床資料預測結果 51
5.4.2 臨床資料加入CT、CTA資訊預測結果 52
5.4.2.1 加入CNN預測結果 52
5.4.2.2 擷取CNN預測結果的前一神經層 54
5.4.3 醫師挑選15項參數預測結果 56
5.4.3.1 加入CNN預測結果 57
5.4.3.2 擷取CNN預測結果的前一神經層 60
5.5 排除手術較差患者與腦出血患者 62
5.5.1 純臨床資料預測結果 62
5.5.2 臨床資料加入CT、CTA資訊預測結果 63
5.5.2.1 加入CNN預測結果 64
5.5.2.2 擷取CNN預測結果的前一神經層 66
5.5.3 醫師挑選15項參數預測結果 68
5.5.3.1 加入CNN預測結果 68
5.5.3.2 擷取CNN預測結果的前一神經層 71
5.6 實驗討論 73
第六章　結論與未來展望 75
6.1 結論 75
6.2 未來展望 76
參考文獻 77

[1]"The top 10 causes of death", WHO, 2020-12-09，取自: https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death.
[2]"Long working hours increasing deaths from heart disease and stroke: WHO, ILO", WHO, 取自: https://www.who.int/news/item/17-05-2021-long-working-hours-increasing-deaths-from-heart-disease-and-stroke-who-ilo.
[3]"2021世界中風日 「腦中風紅色警報」評估表揪出腦中風潛在高危險族！"，衛生福利部新聞，2021-10-19，取自: https://www.mohw.gov.tw/cp-5021-63690-1.html.
[4]Hsieh FI, Chiou HY. "Stroke: morbidity, risk factors, and care in taiwan", J Stroke, 2014 May;16(2):59-64.
[5]Hsieh FI, Lien LM, Chen ST, et al. "Get With the Guidelines-Stroke performance indicators: surveillance of stroke care in the Taiwan Stroke Registry: Get With the Guidelines-Stroke in Taiwan", Circulation 2010;122:1116-23.
[6]Southerland AM. "Clinical Evaluation of the Patient With Acute Stroke", Continuum (Minneap Minn) 2017;23:40-61.
[7]Grotta JC, Savitz SI, Persse D. "Stroke severity as well as time should determine stroke patient triage", Stroke 2013;44:555-7.
[8]Nilsson, Nils John. "The Quest for Artificial Intelligence: A History of Ideas and Achievements", 2010.
[9]Angela Spatharou, Solveigh Hieronimus, and Jonathan Jenkins. "Transforming healthcare with AI: The impact on the workforce and organizations", Executive Briefing, March 2020.
[10]圖片來源：https://www.ndc.gov.tw/Content_List.aspx?n=695E69E28C6AC7F3, Apr 2022.
[11]高明見，"腦中風預防篩檢的重要性：運用人工智慧(AI)技術篩檢中風高風險族群"，Taiwan Medical Journal, Vol.63, pp.16-20, 2020.
[12]L.M. Prevedello, B.S. Erdal, J.L. Ryu, K.J. Little, M. Demirer, S. Qian, et al. "Automated critical test findings identification and online notification system using artificial intelligence in imaging", Radiology, Vol. 285, pp. 923-931, 2017.
[13]Nathan A. Shlobin, Ammad A. Baig, Muhammad Waqas, et al. "Artificial Intelligence for Large-Vessel Occlusion Stroke: A Systematic Review", World Neurosurgery, Vol 159, pp. 207-220, 2022.
[14]Xie Y, Jiang B, Gong E, et al. "Use of gradient boosting machine learning to predict patient outcome in acute ischemic stroke on the basis of imaging, demographic, and clinical information", American Journal of Roentgenology, Vol 212, pp. 44-51, 2019.
[15]Lyden P. "Using the National Institutes of Health Stroke Scale: A Cautionary Tale. Stroke”, Stroke, Vol 48(2), pp. 513-519, 2017.
[16]NINDS rt-PA Stroke Study Group. "Tissue plasminogen activator for acute ischemic stroke.", N Engl J Med, Vol 333, pp. 1581–1587, 1995.
[17]湯頌君，"缺血性腦中風與靜脈血栓溶解治療"，臺大醫院神經部，2009.
[18]張谷州，"就要滿十年了！回頭看看IV rt-PA治療缺血性腦中風的規定（之一）"，社團法人台灣腦中風學會，2018.
[19]Mehta, Rhea , Chinthapalli, et al. "Glasgow coma scale explained", British Medical Journal, Vol 365, 2019.
[20]Enriquez CM, Chisholm KH, Madden LK, et al. "Glasgow Coma Scale: Generating Clinical Standards", Journal of Neuroscience Nursing, vol 51(3), pp. 142-146, 2019.
[21]Hulin Kuang, Bijoy K. Menon, Sung IL Sohn, et al. "EIS-Net: Segmenting early infarct and scoring ASPECTS simultaneously on non-contrast CT of patients with acute ischemic stroke", Medical Image Analysis, Vol 70, 2021.
[22]Dargazanli C, Fahed R, Blanc R, et al. "Modi¬ed Thrombolysis in Cerebral Infarction 2C/Thrombolysis in Cerebral Infarction 3 Reperfusion Should Be the Aim of Mechanical Thrombectomy", Insights From the ASTER Trial (Contact Aspiration Versus Stent Retriever for Successful Revascularization), Stroke, Vol 49, 2018.
[23]Thomas M Cover, Peter E Hart. "Nearest neighbor pattern classification", IEEE Transactions on Information Theory, 1967.
[24]Hunt, E B, Marin, J, & Stone, P J. "Experiments in induction", Academic Press, 1966.
[25]Tin Kam Ho, "Random decision forests," Proceedings of 3rd International Conference on Document Analysis and Recognition, Vol 1, pp. 278-282, 1995.
[26]Cortes C, Vapnik V. "Support-vector networks", Machine Learning, Vol 20, pp. 273–297, 1995.
[27]C. Seiffert, T. M. Khoshgoftaar, J. Van Hulse, et al. "RUSBoost: Improving classification performance when training data is skewed", 2008 19th International Conference on Pattern Recognition, pp. 1-4, 2008.
[28]Freund, Yoav, Schapire, Robert E. "A Decision-Theoretic Generalization of on-Line Learning and an Application to Boosting", CiteSeerX, Vol 10, 1995.
[29]Hancock JT, Khoshgoftaar TM. "CatBoost for big data: an interdisciplinary review", J Big Data. Vol 7(1):94, 2020.
[30]Cheng Li. "A Gentle Introduction to Gradient Boosting", College of Computer and Information Science Northeastern University, Nov 2019.
[31]圖片來源：https://scikit-learn.org/stable/modules/cross_validation.html, May, 2022.
[32]Fukushima, K. "Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position", Biological Cybernetics Vol 36, pp. 193–202, 1980.
[33]Springenberg Jost Tobias, Dosovitskiy Alexey, Brox, Thomas, et al. "Striving for Simplicity: The All Convolutional Net", Machine Learning, Apr, 2015.
[34]Y. Lecun, L. Bottou, Y. Bengio, et al. "Gradient-based learning applied to document recognition", Proceedings of the IEEE, Vol 86, pp. 2278-2324, 1998.
[35]He Kaiming, Zhang Xiangyu, Ren Shaoqing, et al. "Deep Residual Learning for Image Recognition", Computer Vision and Pattern Recognition, pp. 770-778, 2016.
[36]C. Szegedy et al. "Going deeper with convolutions," 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1-9, 2015.
[37]Szegedy, Christian, et al. "Rethinking the inception architecture for computer vision." Computer Vision and Pattern Recognition, pp. 2818-2826, 2016.

[38]Breiman, Leo. "Bagging predictors", Machine learning, pp. 123-140, 1996.
[39]N. V. Chawla, K. W. Bowyer, L. O. Hall, et al. "SMOTE: synthetic minority over-sampling technique." Artificial Intelligence, pp. 321-357, 2002.
[40]Alberto Fernández, Salvador García, Francisco Herrera, and Nitesh V. Chawla. "SMOTE for learning from imbalanced data: progress and challenges, marking the 15-year anniversary" Journal of Artificial Intelligence Research, Vol 61, pp. 863–905, 2018.
[41]Ebrahimi Shahabadi, Mohammad Saleh, Tabrizchi Hamed, Kuchaki Rafsanjani, et al. "A combination of clustering-based under-sampling with ensemble methods for solving imbalanced class problem in intelligent systems", Technological Forecasting and Social Change, Elsevier, Vol 169, 2021.
[42]圖片來源：https://imbalanced-learn.org/dev/auto_examples/under-sampling/plot_illustration_nearmiss.html#sphx-glr-auto-examples-under-sampling-plot-illustration-nearmiss-py, April, 2022.
[43]Sukirty Jain, Sanyam Shukla, Rajesh Wadhvani. "Dynamic selection of normalization techniques using data complexity measures", Expert Systems with Applications. Vol 106, pp. 252-262, 2018.