深度學習已經成功應用在許多醫學影像上，例如肺癌、膀胱癌、乳腺癌、血管鈣化分析等等，在影像判讀上與脈診相似的心電圖，都是由波形組成，並在波形中定義各種特徵從而判斷健康程度或是有無其他疾病等。在本篇論文中，提出了卷積神經網路模型來辨識滑脈以及弦脈這兩個最常見的脈象，經過實驗後，可以精確分別兩種脈象，在簡單的CNN網路中可獲得91.79%的準確度，在更複雜的網路模型中更達到95.35%。並透過熱力圖說明這些模型的可信度。

Deep learning has been successfully applied to many medical images, such as lung cancer, bladder cancer, breast cancer, vascular calcification analysis, etc. The ECG(Electrocardiography), which is similar to pulse diagnosis in image interpretation, both of them are composed of waveforms. We can judge the degree of health or influenced by some diseases by defining various features in the waveforms. In this paper, we proposed a convolutional neural network model to identify the two most common pulse: slippery pulse and wiry pulse. The thesis model has 91.79% accuracy, and with complicated model we reached 95.35%. In the end, we illustrated the credibility of the models with heatmap.

第一章 緒論 1
1.1研究背景 1
1.2研究目的 2
1.3論文架構 2
第二章 相關文獻探討 4
2.1醫學影像在深度學習的應用 4
2.1.1醫學相關文獻 4
2.1.2脈診相關研究 4
2.2中醫介紹 5
2.2.1傳統中醫介紹 5
2.2.2脈波波形定義 6
2.2.3弦脈 7
2.2.4滑脈 8
2.3小波轉換 8
2.3.1離散小波轉換 9
2.3.2逆離散小波轉換 9
2.4卷積神經網路 11
2.5混淆矩陣(Confusion Matrix) 12
第三章 研究方法 15
3.1資料集 15
3.2數據預處理 17
3.2.1去除雜訊 17
3.2.2單一波形切割（Segmentation） 18
3.3模型建立與訓練 20
第四章 實驗結果 24
4.1神經網路 24
4.2模型可視化解釋 28
第五章 結論與未來展望 32
參考文獻 33

[1]Khosravi P, Kazemi E, Imielinski M, Elemento O, Hajirasouliha I. Deep Convolutional Neural Networks Enable Discrimination of Heterogeneous Digital Pathology Images. EBioMedicine. 2018 Jan;27:317-328. doi: 10.1016/j.ebiom.2017.12.026. Epub 2017 Dec 28. PMID: 29292031; PMCID: PMC5828543.
[2]Vesal, S., Ravikumar, N., Davari, A.A., Ellmann, S., & Maier, A. (2018). Classification of Breast Cancer Histology Images Using Transfer Learning. In ICIAR 2018: Image Analysis and Recognition (pp. 812--819). Springer. doi: 10.1007/978-3-319-93000-8_92
[3]Zhou FY, Jin LP, Dong J. Premature ventricular contraction detection combining deep neural networks and rules inference. Artif Intell Med. 2017 Jun;79:42-51. doi: 10.1016/j.artmed.2017.06.004. Epub 2017 Jun 9. PMID: 28662816.
[4]X. Hu, H. Zhu, J. Xu, D. Xu and J. Dong, "Wrist pulse signals analysis based on Deep Convolutional Neural Networks," 2014 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology, 2014, pp. 1-7, doi: 10.1109/CIBCB.2014.6845525.
[5]Li, G., Watanabe, K., Anzai, H. et al. Pulse-Wave-Pattern Classification with a Convolutional Neural Network. Sci Rep 9, 14930 (2019).
[6]H. Wang, "A Quantitative Method for Pulse Strength Classification Based on Decision Tree," 2008 International Symposium on Information Science and Engineering, 2008, pp. 111-115, doi: 10.1109/ISISE.2008.164.
[7]Hongxia, G., & Binghe, W. (2006). Recognition Method of TCM Pulse-Conditions Based on Wavelet Packet Analysis and BP Neural Networks. Application Research of Computers.
[8]C. Huo, Y. Shi and Y. Zhang, "System for Recognition of Chinese Medical Pulse Signal Based on Wavelet Analysis and Back Propagation Neural Network," 2008 3rd International Conference on Innovative Computing Information and Control, 2008, pp. 126-126, doi: 10.1109/ICICIC.2008.531.
[9]S. Zhang and Q. Sun, "Human pulse recognition based on wavelet transform and BP network," 2015 IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC), 2015, pp. 1-4, doi: 10.1109/ICSPCC.2015.7338973.
[10]S. Zhang and Q. Sun, "Human Pulse Recognition Based on Convolutional Neural Networks," 2016 International Symposium on Computer, Consumer and Control (IS3C), 2016, pp. 366-369, doi: 10.1109/IS3C.2016.101.
[11]B. H. Wang ,J. L. Xian,.“Fuzzy Cluster of Human Body Pulse SignalsBased on AR Model” (in Chinese)., Application Acoustics, 3(2001), pp.23-28.
[12]費兆馥(2003),現代中醫脈診學 (pp. 162-351),人民衛生出版社。
[13]神戶中醫學研究會(1996),舌診與脈診 (pp. 76-134),旺文社。
[14]周松華、朱德增,「弦脈的血流動力學機理探討」,遼寧中醫雜誌, 21(2):49-50,1994.
[15]張葉青、董耀榮、王憶勤、燕海霞、許文杰、許朝霞、郭睿、劉國萍、郝一鳴、錢鵬,「高血壓病弦脈和非高血壓病弦脈脈圖參數的比較」世界科學技術:中醫藥現代化,1:1302-1305,2012。
[16]Y. LeCun et al., "Deep learning," nature, vol. 521, no. 7553, p. 436, 2015.
[17]Y. Lecun, L. Bottou, Y. Bengio and P. Haffner, "Gradient-based learning applied to document recognition," in Proceedings of the IEEE, vol. 86, no. 11, pp. 2278-2324, Nov. 1998, doi: 10.1109/5.726791.
[18]Lisheng Xu, D. Zhang and Kuanquan Wang, "Wavelet-based cascaded adaptive filter for removing baseline drift in pulse waveforms," in IEEE Transactions on Biomedical Engineering, vol. 52, no. 11, pp. 1973-1975, Nov. 2005, doi: 10.1109/TBME.2005.856296.
[19]C. Xia et al., "A Practical Approach to Wrist Pulse Segmentation and Single-period Average Waveform Estimation," 2008 International Conference on BioMedical Engineering and Informatics, 2008, pp. 334-338, doi: 10.1109/BMEI.2008.140.
[20]Yinghui Chen, Lei Zhang, David Zhang, Dongyu Zhang,Wrist pulse signal diagnosis using modified Gaussian models and Fuzzy C-Means classification,Medical Engineering & Physics,Volume 31, Issue 10,2009,Pages 1283-1289,ISSN 1350-4533,https://doi.org/10.1016/j.medengphy.2009.08.008.
[21]F. Chollet, “Xception: Deep learning with depthwise separable convolutions,” CoRR, vol. abs/1610.02357, 2016.
[22]C. Szegedy, S. Ioffe, V. Vanhoucke, and A. Alemi, “Inception-v4, inception-resnet and the impact of residual connections on learning,” 2016.
[23]Ulas Baran Baloglu, Muhammed Talo, Ozal Yildirim, Ru San Tan, U Rajendra Acharya,Classification of myocardial infarction with multi-lead ECG signals and deep CNN,Pattern Recognition Letters,Volume 122,2019,Pages 23-30,ISSN 0167-8655,https://doi.org/10.1016/j.patrec.2019.02.016.
[24]R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-cam: Visual explanations from deep networks via gradient-based localization,” in Proceedings of the IEEE international conference on computer vision, pp. 618–626, 2017.