近年來許多人透過醫學美容滿足對外貌的期望，根據ASAPS統計顯示，雷射除毛在醫學美容的雷射治療項目中名列第一，每年平均百萬人次以上。傳統的雷射除毛一般採用較大的光斑(直徑6~18 mm)，且醫師需配戴護目鏡操作雷射手持裝置，通常不易透過其肉眼觀察除毛治療前後的情形。目前醫學影像科技已日趨成熟且應用廣泛，本研究目的為了配合創新影像導向的雷射除毛方式，設計發展皮膚毛孔偵測系統。透過系統將皮膚影像放大顯示，能讓醫師立即清楚觀察到雷射治療時的皮膚狀況；此外，也可以利用影像偵測定位毛孔，未來搭配雷射小光斑(直徑1 mm)的飛點掃描進行除毛。其特定目標為：(1) 皮膚影像擷取裝置設計；(2) 毛孔偵測之影像處理方法；(3) 系統驗證分析。

本研究考量皮膚毛孔偵測系統的臨床需求與功能性設計，設計與完成了(a) 皮膚影像擷取裝置：手持裝置固定輔助性光源、攝影機與鏡頭，傳輸影像訊號到電腦；(b) 分析皮膚毛孔影像的特性：直方圖和特徵邊緣強度剖面分析；(c) 毛孔偵測之影像處理方法：皮膚ROI影像經過遮罩運算，調整適當的門檻值分割為二值化影像，接著形態學的閉合運算，再以連通標記法得出各區域，調整適當的最小像素總數篩選為毛孔特徵，最後定位毛孔。(d) 系統驗證分析：建立標準影像，利用統計學靈敏度、特異性及ROC曲線，驗證系統的性能。

經由系統偵測130張皮膚影像，分析門檻值參數(120、150和180)及最小像素總數參數(10、20和30)其靈敏度與特異性，並做ROC曲線統計。結果顯示以門檻值180的AUC值(= 0.895)最大，其對無辨別率線最遠點的最小像素總數介於10到20之間，其可做為皮膚毛孔偵測系統的較佳參數組合。

本研究未來可繼續改進的部分，包括：(1) 提高影像解析度而提升系統對於較細毛髮的偵測率。(2) 透過優化程式或利用多執行緒，提高系統執行效率。(3) 經由專業皮膚科醫師建立標準影像，增加受測者與膚色類型(Ⅰ到Ⅵ)之皮膚影像樣本數，系統驗證會更具公信力。(4) 皮膚毛孔影像偵測系統實際結合雷射小光斑飛點掃描，應用於臨床雷射除毛上。

In recent years, people have satisfied expectation of personal appearance through cosmetic surgeries. According to statistics of ASAPS, laser hair removal has been the most applied laser treatment in cosmetic surgeries with the annual average of more than one million procedures. Common laser hair removal is generally treated with big spot size (diameter 6~18 mm). While the clinician securely operates hand piece with a pair of goggles to process hair removal, it is not easy to observe the situations clearly before and after the treatment. Currently, the medical imaging technology has been fully developed and widely applied. The purpose of this study is to design and develop skin pores detection system for the innovative image-oriented laser hair removal treatment. The clinician can clearly observe skin conditions after treatment through the system with magnified skin image. In addition, the system can detect the position of pores by image processing. In the future, the system will be combined with small spot size (diameter 1 mm) laser scanning control module to guide precision laser hair removal. The specific aims are: (1) Design of the image acquisition device; (2) Programming of image processing system for pores detection; (3) System validation and analysis.

To consider the clinical requirements and functional design for skin pores detection system, this study designed and accomplished (a) The image acquisition device: configures the hand piece which includes an auxiliary light source, a camera and lens. The image is then transmitted to the computer. (b) Characteristic analysis of the skin pores image: histogram and the characteristic edge intensity profile. (c) Programming of image processing for pores detection: the skin ROI image is first processed by mask operator. The image with the appropriate threshold value is then divided into a binary image. The binary image is computed by morphological closing operator and possible pores location areas are obtained by connected component labeling. Pores features are selected through adjustment of the appropriate minimum total number of pixels. Finally, pores are marked on the image. (d) System validation and analysis: detect pre-established standard images then utilize statistical sensitivity, specificity and ROC curves to verify the system performance.

130 skin images are detected by the system and analyzed with the gray level threshold (120, 150 and 180) and the minimum total number of pixels (10, 20 and 30). With the obtained sensitivity and specificity, ROC curves are drawn. Results showed that the threshold value which is 180 has the largest AUC value (= 0.895). The minimum total number of pixels which is the farthest point to the line of no discrimination is between 10 and 20. These parameters have been proven to enhance the system performance.

In the future, further improvements of this system includes: (1) To improve detection rate of the thinner hair by increasing the image resolution. (2) To improve system efficiency by the optimum program and utilizable multi-threaded. (3) To expand the standard image gallery by adding skin types (Ⅰ~Ⅵ) samples through a professional dermatologist, the system validation will be more credible. (4) To be combined with small spot size (diameter 1 mm) laser scanning control module for guiding precision laser hair removal.

[1] 美國整形美容外科協會(ASAPS) http://www.surgery.org/media/statistics
[2] Leon Goldman, Donald J. Blaney, Alan Freemond, The Biomedical Aspects of Lasers, JAMA. Vol. 188(3) : 302-306,1964
[3] R. R. Anderson, J. A. Parish, Selective photothermolysis:Precise microsurgery by selective absorption of pulsed radiation, Secience, Vol. 220：524-527
[4] 許士昌，解剖生理學，永大書局有限公司，2009
[5] 台大皮膚科部，實用皮膚醫學(第二版)，金名，2006
[6] Elaine N. Marieb, Jon Mallatt, Patricia Brady Wilhelm, Human Anatomy, 5th,
Benjamin Cummings, 2007
[7] Mitchel P. Goldman, Robert A. Weiss, etc, Advabced techiniques in
dermatologic surgery, Francis, 2006
[8] Murad Alam, Ashish C. Bhatia, Roopal V. Kundu, Simon S. Yoo, Henry Hin-Lee Chan, Cosmetic Dermatology for Skin of Color, McGraw-Hill, 2009
[9] 楊燕文，毛囊再生能力及其影響因素，中國美容醫學, Vol.15(12): 1415-1419, 2006
[10] George Cotsarelis, Tung-Tien Sun, Robert M. Lavker, Label-retaining cells reside in the bulge area of pilosebaceous unit: Implications for follicular stem cells, hair cycle, and skin carcinogenesis, Cell, Vol. 61(7): 1329-1337, 1990
[11] Mitchel P. Goldman, Cutaneous and Cosmetic Laser Surgery, ELSVIER, 2006
[12] G.B. Altshuler, R.R. Anderson, D. Manstein, H.H. Zenzie, and M.Z. Smirnov, Extended Theory of Selective Photothermolysis, Lasers in Surgery and Medicine 29, 2001
[13] 原著 宮地良樹，古江增隆，松永佳世子，譯者 曾銘儀，醫學美容皮膚科
學 :皮膚科醫師第一本要讀的書，合記，2009
[14] U.S Food and Drug Administration http://www.fda.gov/
[15] 原著 Gonzalez. Woods, 譯者 謬紹剛, Digital Image Processing 3/e, 台灣培生教育，2009
[16] 原著Peter Armstrong, Martin L.Wastie, 譯者 鄭慶明，醫學影像診斷學，合記，2001
[17] 劉杏，眼科臨床光學相干斷層成像學，廣東科技，2006
[18] 鍾國亮，影像處理與機器視覺(第三版)，東華圖書，2006
[19] 求是科技，Visual C++數位影像處理技術大全，文魁資訊，2008
[20] A. V. Oppenheim, R. W. Schafer, Discrete-Time Signal Processing, Prentice Hall, pp. 140-210, 1999
[21] 陳順宇、鄭碧娥，統計學(第三版)，華泰書局，1998
[22] 盧誌銘、藍守仁，接收運作指標(ROC)曲線，台灣醫學，第四十卷，第一
期pp.33-35，1997
[23] 黃欣慧，影像處理技術應用於人臉膚質特徵之研究，元智大學碩士論文，
2005
[24] Qian Zhang, TaegKeun Whangbo, Skin Pores Detection for Image-Based Skin Analysis, IDEAL, pp. 233–240, 2008
[25] Oana G. Cula, Kristin J. Dana, Skin Texture Modeling, Computer Vision,pp.97-119, 2005
[26] M. Chantler, Why illuminant direction is fundamental to texture analysis, IEEE Proc. Vision Image Signal Process, Vol.142：199-206, 1995
[27] Nina Otberg, Heike Richter, Hans Schaefer, Ulrike Blume-Peytavi, Wolfram Sterry, and Jurgen Lademann, Variations of Hair Follicle Size and Distribution in Different Body Sites, the Journal of Investigative Dermatology, Vol. 122：14-19, 2004
[28] Smart Girl Consulting Permanent Facial Hair Removal
http://smartgirlconsulting.com/