頭皮問題成因千百種，其中又以落髮佔頭皮問題大多數比例，其中
落髮問題檢測又可分為侵入式與非侵入式頭皮評估。侵入式頭皮評估
需要進行活體切片，而切片會造成疼痛與傷口的產生，非侵入式頭皮評
估則不會有上述問題。但是評估結果會因醫師及頭皮理療師當下精神
狀態所影響。
因此本研究將著重於非侵入式落髮檢測系統開發，以深度學習架
構進行設計模型並利用階層級聯式實例分割進行高準確度頭髮及毛囊
的分割，接著將分割結果利用電腦視覺計算頭髮密度與寬度，並依據醫
師及頭皮理療師的經驗下建立落髮評分診斷落髮嚴重度。
最後將落髮檢測系統架設於雲端伺服器，減少醫療院所設置高效
能運算主機的必要性。本研究提出的深度學習網路比起其他文獻相比，
有達到較高的準確度。

There are thousands of causes of scalp problems, of which hair loss
accounts for the majority of scalp problems. The detection of hair loss
problems can be divided into invasive and non-invasive scalp assessments.
Invasive scalp assessment requires the biopsy, which can cause pain and
wounding, while non-invasive scalp assessment does not. However, the
results of the assessment will be affected by the current state of mind of the
physician and scalp physiotherapist.
Therefore, this research will focus on the development of a noninvasive
hair loss detection system, design a model with a deep learning
architecture, and use hierarchical cascaded instance segmentation to segment
hair and hair follicles with high accuracy, and then use computer vision to
calculate hair density and hair follicles. According to the experience of
physicians and scalp physiotherapists, a hair loss score was established to
diagnose the severity of hair loss.
Finally, the hair loss detection system is set up on the cloud server to
reduce the necessity of setting up high-performance computing hosts in
medical institutions. Compared with other literatures, the deep learning
network proposed in this study achieves higher accuracy.

[1] R . Rajput, “Understanding hair loss due to air pollution and the approach to management,” Hair Ther Transplant , 2015.
[2] J. Su, L. Chen, C. Hsu, W. Wang, C. Kuo, W. Chang, W. Hu and D. Lee, “An Intelligent Scalp Inspection and Diagnosis System for Caring Hairy Scalp Health,” 2018 IEEE 7th Global Conference on Consumer Electronics (GCCE), pp. 507-508, 2018.
[3] P. Rattanakaemakorn and P. Suchonwanit, “Scalp pruritus: review of the pathogenesis, diagnosis and management,” BioMed research international, 2019.
[4] A. Herbert and P. Honig, “Scaling scalp diseases in children,” Clinical Pediatrics, pp. 374-377, 1983.
[5] R. Dhurat and P. Saraogi, “Hair evaluation methods: merits and demerits,” International journal of trichology, 2009.
[6] R. Hoffmann, “TrichoScan: a novel tool for the analysis of hair growth in vivo,” Journal of Investigative Dermatology Symposium Proceedings, vol.8, no. 1, pp. 109-115, June 2003.
[7] H. Shih, “An Unsupervised Hair Segmentation and Counting System in Microscopy Images,” IEEE Sensors Journal, vol. 15, no. 6, pp. 3565-3572, June 2015.
[8] G. Urban, N. Feil, E. Csuka, K. Hashemi, C. Ekelem, F. Choi, N. Mesinkovska and P. Baldi, “Combining Deep Learning With Optical Coherence Tomography Imaging to Determine Scalp Hair and Follicle Counts,” Lasers Surg Med. pp. 171-178, Jan 2021.
[9] K. Chen, J. Pang, J. Wang, Y. Xiong, X. Li, S. Sun, W. Feng, Z. Liu, J. Shi, W. Ouyang, C. Loy and D. Lin, “Hybrid Task Cascade for Instance Segmentation,” 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 4969 - 4978, 2019.
[10] H. Kang, T. Kang, S. Lee, Y. Park, H. Kim and S. Kim, “The changing patterns of hair density and thickness in South Korean women with hair loss: clinical office-based phototrichogram analysis,” Int J Dermatol, Jan 2009.
[11] E. Ludwig, “Classification of the types of androgenetic alopecia (common baldness) occurring in the female sex,” Br J Dermatol, Sep 1977.
[12] K. Gordon and A. Tosti, “Alopecia: evaluation and treatment,” Clin Cosmet Investig Dermatol, 2011.
[13] R. Toncić, J. Lipozencić and Z. Pastar, “Videodermoscopy in the evaluation of hair and scalp disorders,” Acta Dermatovenerol Croat, 2007.
[14] S. Inui, T. Nakajima and S. Itami, “Scalp dermoscopy of androgenetic alopecia in Asian people,” J Dermatol, Feb 2009.
[15] A. Ishino, M. Uzuka, Y. Tsuji, J. Nakanishi, N. Hanzawa and S. Imamura, “Progressive decrease in hair diameter in Japanese with male pattern baldness,” J Dermatol, Dec 1997.
[16] A. Chaichana, E. Frey, A. Teyateeti, K. Rhoongsittichai, C. Tocharoenchai, P. Pusuwan and K. Jangpatarapongsa, “Automated segmentation of lung, liver, and liver tumors from Tc-99m MAA SPECT/CT images for Y-90 radioembolization using convolutional neural networks,” Med Phys, pp. 7877-7890, 2021.
[17] Y. Lei, Y. Fu, J. Roper, K. Higgins, J. Bradley, W. Curran, T. Liu and X. Yang, “Echocardiographic image multi-structure segmentation using Cardiac-SegNet,” Med Phys, pp. 2426-2437, May 2021.
[18] Z. Liu, H. Yuan and H. Wang, “CAM-Wnet: An effective solution for accurate pulmonary embolism segmentation,” Med Phys, 2022.
[19] K. He, G. Gkioxari, P. Dollár and R. Girshick, “Mask R-CNN,” 2017 IEEE International Conference on Computer Vision (ICCV), pp. 2980-2988, 2017.
[20] S. Ren, K. He, R. Girshick and J. Sun, “Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1137-1149, June 2017.
[21] Z. Cai and N. Vasconcelos, “Cascade R-CNN: High Quality Object Detection and Instance Segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 5, pp. 1483-1498, May 2021.
[22] K. He, X. Zhang, S. Ren and J. Sun, “Deep Residual Learning for Image Recognition,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 770-778, 2016.
[23] T. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan and S. Belongie, “Feature Pyramid Networks for Object Detection,” 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 936-944, 2017.
[24] Z. Cai and N. Vasconcelos, “Cascade R-CNN: High Quality Object Detection and Instance Segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 5, pp. 1483-1498, May 2021.
[25] Z. Zhang, S. Fidler and R. Urtasun, “Instance-Level Segmentation for Autonomous Driving with Deep Densely Connected MRFs,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 669-677, 2016.
[26] A. Arnab and H. Philip, “Bottom-up instance segmentation using deep higher-order crfs,” arXiv:1609.02583, 2016.
[27] Z. Zheng, P. Wang, D. Ren, W. Liu, R. Ye, Q. Hu and W. Zuo, “Enhancing Geometric Factors in Model Learning and Inference for Object Detection and Instance Segmentation,” IEEE Transactions on Cybernetics.
[28] Z. Zheng, P. Wang, W. Liu, J. Li, R. Ye and D. Ren, “Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression,” Proceedings of the AAAI Conference on Artificial Intelligence, pp. 12993-13000, 2020.
[29] R. Bacon, H. Mizoguchi and J. Schwartz, “Assessing therapeutic effectiveness of scalp treatments for dandruff and seborrheic dermatitis, part 1: a reliable and relevant method based on the adherent scalp flaking score (ASFS),” Journal of Dermatological Treatment, vol. 25, pp. 232-236, 2014.
[30] A. Paszke, S. Gross, S. Chintala, G. Chanan, E. Yang, Z. DeVito, Z. Lin, A. Desmaison, L. Antiga and A. Lerer, “Automatic differentiation in pytorch,” In Advances in Neural Information Processing Systems Workshop, 2017.
[31] S. Qiao, L. Chen and A. Yuille, “DetectoRS: Detecting Objects with Recursive Feature Pyramid and Switchable Atrous Convolution,” 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 10208-10219, 2021.