阿茲海默症以及其他失智症目前不但成為全世界最嚴重的問題，也成爲全世界第五大死因。隨著人口老化，台灣社會已呈現高齡化的結構，失智症患者人數不斷地快速增加，造成家庭照顧者的負擔日益嚴峻。相較於腦影像、血液檢查等較高成本的方式，本研究基於神經心理測驗，透過分析長者所描繪的圖片並利用深度學習的方法，建立一個數位自動化阿茲海默症快篩系統。早期偵測阿茲海默症不但能夠提升他們的生活品質，也能夠減輕照顧者的壓力與照護成本。有鑒於此，本研究利用開放手繪資料集來預訓練神經網路，再將所萃取的特徵與學習到的參數進一步建立數位自動化的阿兹海默症快篩系統，輔助臨床診斷。本研究利用了118位長者紙筆描繪的複雜圖形來區分輕度認知障礙與健康長者，透過一系列的實驗進行驗證後，在ROC曲線面積的指標中達到0.913。另外，本研究也蒐集了60位長者利用數位繪圖板來描繪複雜圖形的資料，在區分阿兹海默症與健康長者實驗驗證後，在ROC曲線面積的指標中達到0.950的結果。

Alzheimer’s disease (AD) and the other types of dementia have become one of the most serious global health issues and the fifth leading cause of death worldwide nowadays. Therefore, early detection of the disease in the stage of mild cognitive impairment (MCI), which is a prodromal stage of progressing to AD and mild AD, is crucial in order to improve the quality of life of the patients and to decrease the burden of their caregiver and clinicians. The aim of our study is to design a digital screening system based on the Rey-Osterrieth Complex Figure (ROCF) neuropsychological drawing test in order to assist the clinicians to detect whether the subject is MCI or AD against healthy control (HC) automatically. A data-driven deep learning approach is implemented in this work for building the screening system. An architecture of convolution neural network is designed for pre-training and extracting useful features from the figures drawn by the subjects. The learned features are then transferred to our collected dataset for further training of the classifier in order to distinguish the patients with MCI or AD against HC. As a result, a mean area under the receiver operating characteristic curve score (AUC) of 0.913 is achieved for classifying MCI vs. HC in traditional pencil and paper based ROCF called NTUH_ROCF dataset. On the other hand, dataset that collected using digitalize graphics tablet and smart pen based which is called NTUH_D-ROCF achieved 0.950 of AUC in classifying AD vs. HC.

口試委員會審定書 i
誌謝 ii
中文摘要 iii
ABSTRACT iv
CONTENTS v
LIST OF TABLES xi
Chapter 1 Introduction 1
1.1 Background 1
1.2 Motivation 3
1.3 Challenges 4
1.3.1 Detection of Early Stage of Alzheimer’s Disease 5
1.3.2 High Cost of Manual Features Engineering 5
1.3.3 Efficiency of the Screening Process 6
1.4 Related Work 6
1.4.1 Brain Imaging 7
1.4.2 Neuropsychological Test 8
1.5 Objectives 9
1.5.1 A Screening System Capable of Detecting Early Signs of Alzheimer’s Disease based on Neuropsychological Test 10
1.5.2 A Data-driven and Deep Learning Approach of Alzheimer’s Disease Screening System 10
1.5.3 A Digital and Automated Processing Pipeline 11
1.6 Thesis Organization 12
Chapter 2 Preliminaries 13
2.1 Convolutional Neural Network 13
2.1.1 Network Architecture Overview 13
2.1.2 Convolutional Layer 14
2.1.3 Pooling Layer 15
2.1.4 Fully Connected Layer 16
2.1.5 Training Process of a Neural Network 17
2.2 Transfer Learning 18
2.2.1 Pre-trained Model 19
2.2.2 Pre-trained Model as Features Extractor 19
2.2.3 Fine-tuning the Pre-trained Model 21
2.3 Regularization for Deep Learning 22
2.3.1 Data Augmentation 22
2.3.2 Dropout 23
Chapter 3 Screening System 25
3.1 System Overview 25
3.2 Neuropsychological Test Selection 26
3.3 Digital Device for Neuropsychological Test 27
3.4 Data Collection Procedure 27
3.5 Feature Representation Method 28
3.6 Pre-training Engine 29
3.6.1 TU-Berlin Sketch Dataset 29
3.6.2 Model Architecture 30
3.6.3 Image Data Augmentation Generator 32
3.6.4 Model Training 33
3.7 Alzheimer’s Disease Screening Engine 35
3.7.1 Transfer Learning Architecture 36
3.7.2 Image Data Augmentation Generator 36
3.7.3 Model Training 37
Chapter 4 System Evaluation 38
4.1 Data Preparation 38
4.1.1 NTUH_ROCF Dataset 38
4.1.2 NTUH_D-ROCF Dataset 41
4.2 Performance Metrics 43
4.3 Evaluation Procedure 43
4.4 Evaluation of the NTUH_ROCF Dataset 44
4.4.1 Comparison of the Different ROCF Trials 44
4.4.2 Performance of the Proposed Screening Engine to Classify Mild Cognitive Impairment vs. Healthy Control 46
4.4.3 Effectiveness of the Dropout and Data Augmentation Techniques 47
4.4.4 Comparison of Different Network Architectures 48
4.4.5 Effectiveness of the Transfer Learning Technique 55
4.4.6 Visualization of the Proposed Model 57
4.5 Evaluation of the NTUH_D-ROCF Dataset 58
4.5.1 Comparison of the Different ROCF Trials 58
4.5.2 Performance of the Proposed Screening Engine to Classify Alzheimer’s Disease vs. Healthy Control 59
Chapter 5 Conclusion 61
5.1 Summary 61
5.2 Future Work 62
REFERENCES 63

[1]C. J. A. s. D. I. L. Patterson, UK, "World Alzheimer Report 2018: The State of the Art of Dementia Research: New Frontiers," 2018.
[2]World Health Organization, "The top 10 causes of death," 2018.
[3]J. Gaugler, B. James, T. Johnson, A. Marin, and J. Weuve, "2019 Alzheimer''s disease facts and figures," ALZHEIMERS & DEMENTIA, vol. 15, no. 3, pp. 321-387, 2019.
[4]J. E. Galvin and C. H. Sadowsky, "Practical guidelines for the recognition and diagnosis of dementia," J Am Board Fam Med, vol. 25, no. 3, pp. 367-382, 2012.
[5]A. Kotecha, A. Corrêa, K. Fisher, and J. Rushworth, "Olfactory dysfunction as a global biomarker for sniffing out Alzheimer’s disease: A meta-analysis," Biosensors, vol. 8, no. 2, p. 41, 2018.
[6]Alzheimer''s Society United Against Dementia, "What is mild cognitive impairment (MCI)? : Factsheet 470LP," 2015.
[7]M. F. Folstein, S. E. Folstein, and P. R. McHugh, "“Mini-mental state”: a practical method for grading the cognitive state of patients for the clinician," Journal of psychiatric research, vol. 12, no. 3, pp. 189-198, 1975.
[8]C. P. Hughes, L. Berg, W. Danziger, L. A. Coben, and R. L. Martin, "A new clinical scale for the staging of dementia," The British journal of psychiatry, vol. 140, no. 6, pp. 566-572, 1982.
[9]R. A. Sperling et al., "Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer''s Association workgroups on diagnostic guidelines for Alzheimer''s disease," Alzheimer''s & dementia, vol. 7, no. 3, pp. 280-292, 2011.
[10]George Liao, "MOI: Taiwan officially becomes an aged society with people over 65 years old breaking the 14% mark," ed. Taiwan News, 2018.
[11]Taiwan Alzheimer Disease Association. (2019). An estimation of the population of dementia in Taiwan. Available: http://www.tada2002.org.tw/About/IsntDementia
[12]A. Ward, H. M. Arrighi, S. Michels, and J. M. Cedarbaum, "Mild cognitive impairment: disparity of incidence and prevalence estimates," Alzheimer''s & Dementia, vol. 8, no. 1, pp. 14-21, 2012.
[13]B. Dubois, A. Padovani, P. Scheltens, A. Rossi, and G. Dell’Agnello, "Timely diagnosis for Alzheimer’s disease: a literature review on benefits and challenges," Journal of Alzheimer''s disease, vol. 49, no. 3, pp. 617-631, 2016.
[14]L. Trojano and M. Conson, "Visuospatial and visuoconstructive deficits," Handbook of clinical neurology, vol. 88, pp. 373-391, 2008.
[15]V. Isella et al., "A retrospective survey on rotated drawing in persons with mild cognitive impairment or degenerative dementia," The Clinical Neuropsychologist, vol. 27, no. 8, pp. 1300-1315, 2013.
[16]M. Kasai et al., "Non‐verbal learning is impaired in very mild Alzheimer''s disease (CDR 0.5): Normative data from the learning version of the Rey–Osterrieth Complex Figure Test," vol. 60, no. 2, pp. 139-146, 2006.
[17]A. Collie and P. Maruff, "The neuropsychology of preclinical Alzheimer''s disease and mild cognitive impairment," Neuroscience & Biobehavioral Reviews, vol. 24, no. 3, pp. 365-374, 2000.
[18]N. Bogdanovic, "The Challenges of Diagnosis in Alzheimer''s Disease," US Neurology, vol. 14, no. 1, pp. 15-16, 2018.
[19]N. Schuff and X. J. T. B. j. o. r. Zhu, "Imaging of mild cognitive impairment and early dementia," vol. 80, no. special_issue_2, pp. S109-S114, 2007.
[20]R. J. N. o. a. Sperling, "The potential of functional MRI as a biomarker in early Alzheimer''s disease," vol. 32, pp. S37-S43, 2011.
[21]K. Hu, Y. Wang, K. Chen, L. Hou, and X. J. N. Zhang, "Multi-scale features extraction from baseline structure MRI for MCI patient classification and AD early diagnosis," vol. 175, pp. 132-145, 2016.
[22]E. Challis, P. Hurley, L. Serra, M. Bozzali, S. Oliver, and M. J. N. Cercignani, "Gaussian process classification of Alzheimer''s disease and mild cognitive impairment from resting-state fMRI," vol. 112, pp. 232-243, 2015.
[23]F. Li, L. Tran, K.-H. Thung, S. Ji, D. Shen, and J. Li, "A robust deep model for improved classification of AD/MCI patients," IEEE journal of biomedical and health informatics, vol. 19, no. 5, pp. 1610-1616, 2015.
[24]A. J. J. J. o. p. r. Mitchell, "A meta-analysis of the accuracy of the mini-mental state examination in the detection of dementia and mild cognitive impairment," vol. 43, no. 4, pp. 411-431, 2009.
[25]A. Prange and D. Sonntag, "Modeling cognitive status through automatic scoring of a digital version of the clock drawing test," in Proceedings of the 27th ACM Conference on User Modeling, Adaptation and Personalization, pp. 70-77, 2019.
[26]M. F. Mendez, T. Ala, and K. L. Underwood, "Development of Scoring Criteria for the Clock Drawing Task in Alzheimer''s Disease," Journal of the American Geriatrics Society, vol. 40, no. 11, pp. 1095-1099, 1992.
[27]L. Ehreke, M. Luppa, H.-H. König, and S. Riedel-Heller, "Is the Clock Drawing Test a screening tool for the diagnosis of mild cognitive impairment? A systematic review," International Psychogeriatrics, vol. 22, pp. 56-63, 2009.
[28]J. Myers and K. J. P. A. R. Myers, Odessa, "Rey complex figure test and recognition trial," 1995.
[29]J. Miller et al., "A-12 Screening for Mild Cognitive Impairment (MCI) with the Mini-Mental Status Exam (MMSE) and Rey-Osterrieth Complex Figure Test (ROCF)," vol. 29, no. 6, 2014.
[30]E. Salvadori, F. Dieci, P. Caffarra, and L. J. A. o. C. N. Pantoni, "Qualitative Evaluation of the Immediate Copy of the Rey–Osterrieth Complex Figure: Comparison Between Vascular and Degenerative MCI Patients," vol. 34, no. 1, pp. 14-23, 2018.
[31]R. A. Stern et al., "The Boston qualitative scoring system for the Rey-Osterrieth complex figure: description and interrater reliability," vol. 8, no. 3, pp. 309-322, 1994.
[32]J. H. Bernstein and D. P. Waber, Developmental scoring system for the Rey-Osterrieth complex figure: DSS ROCF. Psychological Assessment Resources, 1996.
[33]Y. LeCun, Y. J. T. h. o. b. t. Bengio, and n. networks, "Convolutional networks for images, speech, and time series," vol. 3361, no. 10, p. 1995, 1995.
[34]V. Nair and G. E. Hinton, "Rectified linear units improve restricted boltzmann machines," in Proceedings of the 27th international conference on machine learning (ICML-10), 2010, pp. 807-814.
[35]Y. A. LeCun, L. Bottou, G. B. Orr, and K.-R. Müller, "Efficient backprop," in Neural networks: Tricks of the trade: Springer, 2012, pp. 9-48.
[36]S. J. Pan and Q. Yang, "A survey on transfer learning," IEEE Transactions on knowledge and data engineering, vol. 22, no. 10, pp. 1345-1359, 2009.
[37]K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv preprint arXiv:1409.1556, 2014.
[38]C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, "Rethinking the inception architecture for computer vision," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 2818-2826.
[39]F. Chollet, "Xception: Deep learning with depthwise separable convolutions," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1251-1258.
[40]K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770-778.
[41]A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," in Advances in neural information processing systems, 2012, pp. 1097-1105.
[42]P. Y. Simard, D. Steinkraus, and J. C. Platt, "Best practices for convolutional neural networks applied to visual document analysis," in Icdar, 2003, vol. 3, no. 2003.
[43]D. Cireşan, U. Meier, and J. Schmidhuber, "Multi-column deep neural networks for image classification," arXiv preprint arXiv:1202.2745, 2012.
[44]I. Sato, H. Nishimura, and K. Yokoi, "Apac: Augmented pattern classification with neural networks," arXiv preprint arXiv:1505.03229, 2015.
[45]R. Wu, S. Yan, Y. Shan, Q. Dang, and G. Sun, "Deep image: Scaling up image recognition," arXiv preprint arXiv:1501.02876, 2015.
[46]J. Sietsma and R. J. Dow, "Creating artificial neural networks that generalize," Neural networks, vol. 4, no. 1, pp. 67-79, 1991.
[47]N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, and R. J. T. J. o. M. L. R. Salakhutdinov, "Dropout: a simple way to prevent neural networks from overfitting," vol. 15, no. 1, pp. 1929-1958, 2014.
[48]M. Y. Chang and A. S. Davis, "Neuropsychological Assessment," in Encyclopedia of Child Behavior and Development, S. Goldstein and J. A. Naglieri, Eds. Boston, MA: Springer US, 2011, pp. 1013-1014.
[49]M.-S. Shin, S.-Y. Park, S.-R. Park, S.-H. Seol, and J. S. Kwon, "Clinical and empirical applications of the Rey–Osterrieth complex figure test," Nature protocols, vol. 1, no. 2, p. 892, 2006.
[50]K. L. Possin, "Visual spatial cognition in neurodegenerative disease," Neurocase, vol. 16, no. 6, pp. 466-487, 2010.
[51]A. Rey, "L''examen psychologique dans les cas d''encéphalopathie traumatique.(Les problems.)," Archives de psychologie, 1941.
[52]P. A. Osterrieth, "Le test de copie d''une figure complexe; contribution a l''etude de la perception et de la memoire," Archives de psychologie, 1944.
[53]WACOM. Available: https://www.wacom.com/en-in/
[54]N. Dalal and B. Triggs, "Histograms of oriented gradients for human detection," in Computer Vision and Pattern Recognition, 2005. CVPR 2005. IEEE Computer Society Conference on, 2005, vol. 1, pp. 886-893: IEEE.
[55]D. G. Lowe, "Distinctive image features from scale-invariant keypoints," International journal of computer vision, vol. 60, no. 2, pp. 91-110, 2004.
[56]H. Bay, T. Tuytelaars, and L. Van Gool, "Surf: Speeded up robust features," in European conference on computer vision, 2006, pp. 404-417: Springer.
[57]G. Csurka, C. Dance, L. Fan, J. Willamowski, and C. Bray, "Visual categorization with bags of keypoints," in Workshop on statistical learning in computer vision, ECCV, 2004, vol. 1, no. 1-22, pp. 1-2: Prague.
[58]S. Belongie, J. Malik, and J. Puzicha, "Shape matching and object recognition using shape contexts," IEEE Transactions on Pattern Analysis & Machine Intelligence, no. 4, pp. 509-522, 2002.
[59]M. Eitz, J. Hays, and M. Alexa, "How do humans sketch objects?," ACM Trans. Graph., vol. 31, no. 4, pp. 44:1-44:10, 2012.
[60]J. Yosinski, J. Clune, Y. Bengio, and H. Lipson, "How transferable are features in deep neural networks?," in Advances in neural information processing systems, 2014, pp. 3320-3328.
[61]Q. Yu, Y. Yang, Y.-Z. Song, T. Xiang, and T. Hospedales, "Sketch-a-net that beats humans," arXiv preprint arXiv:1501.07873, 2015.
[62]X. Glorot and Y. Bengio, "Understanding the difficulty of training deep feedforward neural networks," in Proceedings of the thirteenth international conference on artificial intelligence and statistics, 2010, pp. 249-256.
[63]D. P. Kingma and J. J. a. p. a. Ba, "Adam: A method for stochastic optimization," 2014.
[64]J. Duchi, E. Hazan, and Y. Singer, "Adaptive subgradient methods for online learning and stochastic optimization," Journal of Machine Learning Research, vol. 12, no. Jul, pp. 2121-2159, 2011.
[65]A. J. Jak et al., "Quantification of five neuropsychological approaches to defining mild cognitive impairment," The American Journal of Geriatric Psychiatry, vol. 17, no. 5, pp. 368-375, 2009.
[66]T. N. Tombaugh and N. J. J. J. o. t. A. G. S. McIntyre, "The mini‐mental state examination: a comprehensive review," vol. 40, no. 9, pp. 922-935, 1992.
[67]W.-T. Cheah, W. D. Chang, J.-J. Hwang, S.-Y. Hong, L.-C. Fu, and Y.-L. Chang, "A Screening System for Mild Cognitive Impairment Based on Neuropsychological Drawing Test and Neural Network," in 2019 IEEE International Conference on Systems, Man, and Cybernetics (SMC), 2019: IEEE.
[68]J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, "Imagenet: A large-scale hierarchical image database," in 2009 IEEE conference on computer vision and pattern recognition, 2009, pp. 248-255: Ieee.
[69]G. E. Hinton and R. R. J. s. Salakhutdinov, "Reducing the dimensionality of data with neural networks," vol. 313, no. 5786, pp. 504-507, 2006.
[70]K. Simonyan, A. Vedaldi, and A. Zisserman, "Deep inside convolutional networks: Visualising image classification models and saliency maps," arXiv preprint arXiv:1312.6034, 2013.