肝斑為常見的一種色素沈積的皮膚疾病，其病理特徵為表皮色素增加，黑色素細胞製造黑色素並藉由觸角分送到角質化細胞中。本篇論文提出一個電腦輔助分析肝斑的方法，實驗數據經由醫生將疾病分為四個類別，並提供黑色素細胞觸角特性包括其粗細、結構、影像中的亮度以及生長方向等，結合觀察不同嚴重程度影像後所得到的紋理特徵；並透過數學模型賈伯濾波器量化這些描述於影像中的病理結構，另外，馬可夫隨機場萃取每個像素點的特徵向量以及標籤值，不考慮影像的結構，只根據臨近區域像素值相近的特性分析鄰域系像素以分割影像的紋理特徵。基於這些電腦視覺領域萃取影像特徵的經驗，可以設計初始化卷基層，使神經網路有更好的起始點，並利用醫生分類肝斑疾病四個類別的倍頻影像訓練卷積神經網路，希望藉由結合分析影像結構的賈伯濾波器以及探討像素間關聯性的馬可夫隨機場增進卷基神經網路的表現，使其能準確且快速地分析影像中肝斑疾病的類別。實驗結果顯示此演算法能準確且有效率地分類醫學影像，手動調整初始化卷基層並藉由機器學習演算法有助於節省人力和時間的消耗，並且比起隨機初始化卷基層能有更好的分類表現。

Melasma is a common pigmentation skin disease, and pathological characteristic of Melasma is increased epidermal hyperpigmentation, which produce by Melanocyte and translocate into keratinocyte. This thesis proposes a method of computer-aided diagnosis for analysis of Melasma. The Melasma data is judged into four categories, and provide the characteristics of dendrite including scale, structure, intensity difference and the directions by doctors. Utilizing Gabor filter bank quantify the pathological features which combining the texture phenomenon observed from different categories images and the pathological characteristics above. In addition, Markov random field extract feature vectors and the label in each pixel. Regardless the image structure, analyze the neighbor pixels to segment the image texture based on the similarity of the pixel values in adjacent region. Design the initial kernels of convolution layer according to these feature extraction experience in computer vision field to make the better initial points of convolution neural network. Training convolution neural network by four categories HHGM images of Melasma disease, and try to enhance the performance of convolution neural network via combining the structure analysis with Gabor filter and neighborhood exploration with Markov random field. Hope that it can distinguish the categories of Melasma accurately and quickly in HHG image. Experimental results show that this algorithm can classify medical image accurately and efficiently. Additionally, machine learning algorithm with hand craft initial kernel is contribute to reduce time-consuming and labor-consuming, and have better classification performance than the random initialization convolutional kernel.

[1] C.-K. Sun, "Higher Harmonic Generation Microscopy," in Microscopy Techniques: -/-, J. Rietdorf, Ed., ed Berlin, Heidelberg: Springer Berlin Heidelberg, 2005, pp. 17-56.
[2] S.-Y. Chen, S.-U. Chen, H.-Y. Wu, W.-J. Lee, Y.-H. Liao, and C.-K. Sun, "In vivo virtual biopsy of human skin by using noninvasive higher harmonic generation microscopy," Selected Topics in Quantum Electronics, IEEE Journal of, vol. 16, pp. 478-492, 2010.
[3] J. Gordon Betts, P. Desaix, J. E. Johnson, O. Korol, D. Kruse, B.n Poe, J. Wise, M. D. Womble, K. A. Young “Anatomy and Physiology,” OpenStax College, Rice University, 2013
[4] E. N. Marieb, K. Hoehn, ‘Human Anatomy & Physiology’, Published by Benjamin Cummings - An imprint of Addison Wesley Longman, Inc. 2001
[5] Promocell [online] https://www.promocell.com/about-us/profile/
[6] T. Sikora, "The MPEG-7 visual standard for content description-an overview," Circuits and Systems for Video Technology, IEEE Transactions on, vol. 11, pp. 696-702, 2001.
[7] A. V. Oppenheim, A. S. Willsky, and S. H. Nawab,” Signals and systems,” Prentice Hall, 1997.
[8] B. Osgood, “The Fourier Transform and its Applications,” Electrical Engineering Department, Stanford University, USA.
[9] D. Gabor, "Theory of communication," Journal of the Institution of Electrical Engineers - Part III: Radio and Communication Engineering, vol. 93, pp. 429-457, 1946.
[10] Q. Li, J. You, L. Zhang, and P. Bhattacharya, "A multiscale approach to retinal vessel segmentation using Gabor filters and scale multiplication," Systems, Man and Cybernetics, pp. 3521-3527, 2006.
[11] J. V. Soares, J. J. Leandro, R. M. Cesar, H. F. Jelinek, and M. J. Cree, "Retinal vessel segmentation using the 2-D Gabor wavelet and supervised classification," Medical Imaging, IEEE Transactions, vol. 25, pp. 1214-1222, 2006.
[12] V. S. Vyas, P. Rege , “Automated Texture Analysis with Gabor filter,” GVIP Journal, vol. 6, 2006.
[13] M. Lindenbaum, R. Sandler, “Gabor Filter Analysis for Texture Segmentation,” IEEE, 2006 Conference on Computer Vision and Pattern Recognition Workshop (CVPRW'06), Jun 2006.
[14] B.S. Manjunathi, W.Y. Ma, “Texture Features for Browsing and Retrieval of Image Data,” IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 18, pp. 837-842, 1996.
[15] J. R. Movellan, “Tutorial on Gabor Filters,” 2015. [Online] http://mplab.ucsd.edu/tutorials/pdfs/gabor.pdf
[16] D. Dunn, W.E. Higgins, “Optimal Gabor Filters for Texture Segmentation,” IEEE Transactions on Image Processing, Vol. 4, pp. 947-964, 1995.
[17] J. G. Daugman, "Uncertainty relation for resolution in space, spatial frequency, and orientation optimized by two-dimensional visual cortical filters," JOSA A, vol. 2, pp. 1160-1169, 1985.
[18] A. F. Frangi, W. J. Niessen, K. L. Vincken, and M. A. Viergever, "Multiscale vessel enhancement filtering," Medical Image Computing and Computer-Assisted Interventation—MICCAI’98, ed: Springer, pp. 130-137, 1998.
[19] S. You, E. Bas, D. Erdogmus, and J. Kalpathy-Cramer, "Principal Curved Based Retinal Vessel Segmentation towards Diagnosis of Retinal Diseases," Healthcare Informatics, Imaging and Systems Biology (HISB), 2011 First IEEE International Conference, pp. 331-337, 2011.
[20] T. Lindeberg, Scale-space theory in computer vision: Springer, 1993.
[21] H. Abdi, “The Eigen-Decomposition:Eigenvalues and Eigenvectors,” (2006).
[22] Strang, Gilbert. "Introduction to linear algebra." (2011).
[23] A. Blake, P. Kohli, C. Rother, “Markov random fields for vision and image processing,” (2011).
[24] R. Hu, M.M. Fahmy, “Texture segmentation based on a hierarchical Markov random field model,” IEEE International Sympoisum on Circuits and Systems, Singapore, Singapore.
[25] B.S. Manjunath, R. Chellappa, “Unsupervised texture segmentation using Markov random field models,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 13, pp. 478-482, 1991.
[26] C. M. Grinstead, J. L. Snell, “Introduction to Probability,“ (2006).
[27] R. Kashyap, R. Chellappa, “Estimation and Choice of Neighbors in Spatial-Interaction Models of Images,“ IEEE Transactions on Information Theory, Vol. 29, pp. 60-72, 1983.
[28] C. H. Chen, G. G. Lee, “Image segmentation using multiresolution wavelet analysis and expectation‐maximization (EM) algorithm for digital mammography,” International Journal of Imaging Systems and Technology, Vol. 8, pp. 491-504, 1997.
[29] R. Kindermann, J. L. Snell, “Markov Random Fields and Their Applications,” American Mathematical Society Providence, 1980.
[30] F. Liu, J. Y. Gong, “Image segmentation using energy minimization and Markov random fields,” International Conference on Computer and Management (CAMAN), 2011.
[31] D. Geiger, A. Yuille, “ A common framework for image segmentation,” 10th International Conference on Pattern Recognition, 1990.
[32] I. T. Jolliffe, “Mathematical and Statistical Properties of Population Principal Components, ” Springer Series in Statistics, Springer, pp. 8-22, 1986.
[33] J. A. Hartigan and M. A. Wong, "Algorithm AS 136: A k-means clustering algorithm," Applied statistics, pp. 100-108, 1979.
[34] S. S. Khan, A. Ahmad, “Cluster Center Initialization Method for K-means Algorithm Over Data Sets with Two Clusters,” Pattern Recognition Letters, Vol. 25, pp. 1293-1302, 2004.
[35] P. Baldi , “Autoencoders, Unsupervised Learning, and Deep Architectures,” UTLW'11 Proceedings of the 2011 International Conference on Unsupervised and Transfer Learning workshop - Volume 27, pp. 37-50, 2011
[36] G. E. Hinton, Ri. S. Zemel “Autoencoders, minimum description length and Helmholtz free energy,” NIPS'93 Proceedings of the 6th International Conference on Neural Information Processing Systems, pp. 3-10, 1993.
[37] C. Cortes, V. Vapnik, “Support-Vector Networks,” Machine learning, vol.20, pp. 273-297, 1995.
[38] C. J.C. Burges, “A Tutorial on Support Vector Machines for Pattern Recognition,” Data Mining and Knowledge Discovery, Vol. 2, pp. 121-167, 1998.
[39] T. Kohonen, "An introduction to neural computing," Neural networks, vol. 1, pp. 3-16, 1988.
[40] S. Haykin, “Neural Networks: A Comprehensive Foundation,” (1994)
[41] Lecun, Y. (1989). Generalization and network design strategies. In R. Pfeifer, Z. Schreter, F. Fogelman, & L. Steels (Eds.), Connectionism in perspective. Zurich, Switzerland: Elsevier.
[42] Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner, "Gradient-based learning applied to document recognition," Proceedings of the IEEE, vol. 86, pp. 2278-2324, 1998.
[43] I. Goodfellow, Y. Bengio, A. Courville, “Deep Learning,” MIT press, 2016.
[44] G. Hinton, S. Osindero, et al., "A Fast Learning Algorithm for Deep Belief Nets," Neural Computation, vol. 18, pp. 1527-1554, 2006.
[45] Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. "Imagenet classification with deep convolutional neural networks." Advances in neural information processing systems. 2012.
[46] Szegedy, Christian, et al. "Going deeper with convolutions." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2015.
[47] J. Masci, U. Meier, D. Cireşan, and J. Schmidhuber, "Stacked convolutional auto-encoders for hierarchical feature extraction," presented at the Proceedings of the 21th international conference on Artificial neural networks - Volume Part I, Espoo, Finland, 2011.
[48] X.J. Mao, C. Shen, Y.B. Yang, “Image Restoration Using Convolutional Auto-encoders with Symmetric Skip Connections,” Computer Vision and Pattern Recognition, 2016. [arXiv:1606.08921]
[49] R. Caruana, S. Lawrence, C. L. Giles, “Overfitting in Neural Nets: Backpropagation, Conjugate Gradient, and Early Stopping,” in PROC. NEURAL INFORMATION PROCESSING SYSTEMS CONFERENCE, pp. 402, 2001.
[50] L. Prechelt, “Automatic Early Stopping Using Cross Validation: Quantifying the Criteria,” Neural networks: the official journal of the International Neural Network Society, vol.11, pp. 761-767, 1998.
[51] N. Srivastava, G. Hinton, A. Krizhevsky, I. Sutskever, R. Salakhutdinov, “Dropout: A Simple Way to Prevent Neural Networks from Overtting,” Journal of Machine Learning Research, 2014.
[52] S. J. Pan, Q. Yang, “A Survey on Transfer Learning,” IEEE Transactions on Knowledge and Data Engineering, vol. 22, pp.1345-1359, 2010.
[53] W. Dai, Y. Chen, G. R. Xue, Q. Yang, Y. Yu, “Translated Learning: Transfer Learning across Different Feature Spaces, ” Advances in Neural Information Processing Systems, pp. 353-360, 2008.
[54] S. Gupta, R. Girshick, P. Arbelez, and J. Malik, “Learning rich features from rgb-d images for object detection and segmentation,” European Conference on Computer Vision (ECCV), pp. 345–360, 2014.
[55] S. Gupta, P. Arbelez, R. Girshick, and J. Malik, “Indoor scene understanding with rgb-d images: Bottom-up segmentation, object detection and semantic segmentation,” International Journal of Computer Vision, vol. 112, pp. 133–149, 2015.
[56] A. Gupta, M. Ayhan, and A. Maida, “Natural image bases to represent neuroimaging data,” International Conference on Machine Learning (ICML), pp. 987–994, 2013.
[57] H. Chen, Q. Dou, D. Ni, J. Cheng, J. Qin, S. Li, and P. Heng, “Automatic fetal ultrasound standard plane detection using knowledge transferred recurrent neural networks,” MICCAI, 2015, pp. 507–514.
[58] H. Shin, L. Lu, L. Kim, A. Seff, J. Yao, and R. Summers, “Interleaved text/image deep mining on a large-scale radiology image database,” IEEE Conference: The Computer Vision Foundation (CVPR), pp. 1–10, 2015.
[59] H. C. Shin, H. R. Roth, M. Gao, L. Lu, Z. Xu, I. Nogues, J. Yao, D. Mollura, R. M. Summers, “Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning,” IEEE Transactions on Medical Imaging, vol. 35, pp. 1285-1298, 2016.
[60] S. J. Pan, J. T. Kwok, Q. Yang, “Transfer Learning via Dimensionality Reduction,” Association for the Advancement of Artificial Intelligence (AAAI), pp.677-682, 2008.
[61] R.C. Gonzalez and R.E. Woods, Digital Image Processing: Prentice-Hall, Englewood Cliffs, NJ, 2002.
[62] S. Yuheng, Y. Hao, “Image Segmentation Algorithms Overview,” 2017.
[63] N. Senthilkumaran, S. Vaithegi, “Image Segmentation By Using Thresholding Techniques For Medical Images,” Computer Science & Engineering: An International Journal (CSEIJ), Vol. 6, 2016.
[64] M. Sezgin, B. Sankur, “Survey over image thresholding techniques and quantitative performance evaluation, ” Journal of Electronic Imaging, 2004
[65] N. Otsu, "A threshold selection method from gray-level histograms," Automatica, vol. 11, pp. 23-27, 1975.
[66] H. J. Vala, A. Baxi, “A Review on Otsu Image Segmentation Algorithm, ” International Journal of Advanced Research in Computer Engineering & Technology (IJARCET), Vol. 2, 2013.
[67] K. Hammouda, “Texture Segmentation Using Gabor Filters,” Fourth International Conference on Knowledge-Based Intelligent Engineering Systems and Allied Technologies. Proceedings, 2002.
[68] G. Hinton, S. Osindero, et al., "A Fast Learning Algorithm for Deep Belief Nets," Neural Computation, vol. 18, pp. 1527-1554, 2006.
[69] Bottou, Léon. "Stochastic gradient descent tricks." Neural Networks: Tricks of the Trade. Springer Berlin Heidelberg, 2012. 421-436.
[70] Bishop, Christopher M. "Pattern recognition." Machine Learning 128 (2006).
[71] C. Zhang, Q. Liao, A. Rakhlin, B. Miranda, N. Golowich, T. Poggio, “Theory of Deep Learning IIb: Optimization Properties of SGD, ” CBMM Memos, 2017.
[72] N. Qian, “On the Momentum Term in Gradient Descent Learning Algorithm,” Neural networks: the official journal of the International Neural Network Society, 1999.
[73] G. B. Arfken, H. J. Weber, “MATHEMATICAL METHODS FOR PHYSICISTS,” Academic Press, 6th, 2005.
[74] Y. Jia. Caffe | Deep Learning Framework. Available: http://caffe.berkeleyvision.org/
[75] E.Y. Tay, E.Y. Gan, V.W.D. Tan, Z. Lin, Y. Liang, F. Lin, W. Ser, T.G. Thng, “Pilot study of an automated method to determine melasma area and severity index,” Britsh. Journal of Dermatology, Vol. 172, pp. 1535–1540, 2015.
[76] Y. Liang, L. Sun, W. Ser, F. Lin, E.Y. Tay, E.Y. Gan, T.G. Thng, Z. Lin, “Hybrid threshold optimization between global image and local regions in image segmentation for melasma severity assessment,” Multidimens. Syst. Signal Process, vol. 28, pp. 977–994, 2017.
[77] Y. Liang, L. Sun, W. Ser, F. Lin, S.T.G. Thng, Q. Chen, Z. Lin. “Classification of non-tumorous skin pigmentation disorders using voting based probabilistic linear discriminant analysis,” Computers in Biology and Medicine, vol. 99 pp. 123-132, 2018.
[78] Andre Esteva, Brett Kuprel, Roberto A. Novoa, Justin Ko, Susan M. Swetter, Helen M. Blau, Sebastian Thrun, “Dermatologist-level classification of skin cancer with deep neural networks,” Nature, vol. 542, pp.115-118, 2017.