顱顏界標點是一些可用來表示頭顱解剖結構的特徵點，在齒顎矯正的測顱分析中是一項重要的參考依據。牙醫師利用側面的頭顱測顱片上的界標點可計算出一些相關的數據，例如：距離、角度等，以便評估術前術後與發育的顱顏變化。由於目前擷取界標點的方法極為耗費時間和人力、容易產生人為誤差且依賴醫師本身的經驗。因此本研究應用顱顏型態和粗到細的技術，將描軌圖利用影像處理方法擷取出可代表顱顏型態的特徵參數，然後應用K-means分類法完成特徵參數的分類，再將各類別的影像利用四個界標點(Go, Po, N, Me)，使用最小平方誤差方法將其重疊以得到各類的描軌圖樣版。另外，在測顱片影像的處理上，先以直方圖等化強化影像的強度，並使用拉普拉斯-高斯運算子取得邊緣，再由霍氏轉換取得特徵參數，其中利用影像的二分之ㄧ平均亮度做為閾值，分割影像的硬組織與軟組織部份，以利於特徵之擷取。藉由計算特徵參數之間的歐基里得距離，將描軌圖疊到最短距離的測顱片上，最後在樣版上的界標粗略範圍內，經由局部調整以得到測顱片上的界標位置。本研究使用50張描軌圖影像，共可分為六群，每群分別有4張、8張、10張、7張、13張、8張描軌圖影像。經由測顱片的型態分類與描軌圖疊印，可得到測顱片上界標點(Go, Po, S, N, Me)的粗略範圍。

The craniofacial landmark is the point representing the anatomical feature of the skull. It plays an important role in the cephalometry. The orthdontists use these landmarks obtained from cephalogram to compute the related data such as distances, angles, etc for pre- and post-treatment, and growth analysis. The manual measurement for landmarking is time-consuming, tedious and subjective. Therefore, the purpose of this study is to apply skull morphological information and coarse-to-fine technique for landmarking. The morphological feature parameters of tracings are extracted using image processing. Then, these parameters are classified by K-means clustering method. Afterwards, the template for each class are established based on the four landmarks (Go, Po, N, Me) using least-square-error method. In the cephalogram processing, the image intensity is enhanced by histogram equalization. Then, the edge of image is detected using Laplacian-of-Gaussian operator. So, the same morphological feature parameters as tracings are extracted by Hough Transform. For differentiating the hard and soft tissues, the image is segmented with half intensity of cephalogram. Base on the computation of Euclidean distance of morphological feature parameters, the tracing template with the minimal distance is superimposed on cephalogram using morphological feature parameters as mentioned above. The precise positions of landmarks are then extracted within a circular searching range by the image processing. In this study, fifty tracings are as the data set for classification analysis. They are classified six clusters. Each cluster has twenty-two images, thirteen images, and fifteen images, respectively. The searching ranges for the landmark Go, Po, S, N, and Me are obtained from cephalogram is then processed with morphological classification and tracing template superimposition.

[1] A. Jacobson, Radiographic Cephalometry: From Basics to Videoimaging, Chicago: Quintessence Publishing Co., 1995.
[2] http://www.ortho.org.tw/qa3.htm
[3] Y.-T. Chen, Automatic Analysis of Landmarks in Cephalograms, Master thesis, Institute of Biomedical Engineering, National Cheng Kung University, 1991.
[4] Y.-T. Chen, The Integrated Cephalometric Analysis System, Doctoral dissertation, Institute of Biomedical Engineering, National Cheng Kung University, 2000.
[5] J. Cardillo and M. A. Sid-Ahmed, “An image processing system for locating craniofacial landmarks,” IEEE Trans. Med. Imag., vol.13, no.2, pp.275-289, 1994.
[6] T. J. Hutton, S. Cunningham, and P. Hammond, “An evaluation of active shape models for the automatic identification of cephalometric landmarks,” European Journal of Orthodontics, vol.22, no.5, pp.499-508, 2000.
[7] V. Grau, M. C. Juan, C. Monserrat, and C. Knoll, “Automatic localization of cephalometric landmarks,” Journal of Biomedical Informatics, vol.34, no.3, pp.146-156, 2001.
[8] I. El-Feghi, M. A. Sid-Ahmed, and M. Ahmadi, “Localization of craniofacial landmarks using multi-layer perceptron,” Pattern Recognition, vol.37, no.4, pp.609-621, 2004.
[9] R. C. Gonzalez and R. E. Woods, Digital Image Processing, 2nded, Prentice-Hall Press, 2002.
[10] A. P. Dhawan, Medical Image Analysis, John Wiley & Sons, 2003.
[11] A. D. Kulkarni, Computer Vision and Fuzzy-Neural Systems, Ch.8, pp.357-358, Prentice-Hall, 2001.
[12] L.-H. Chen, The Identification and Clustering Analysis of Auditory Neurons for Salicylate-Induced Rat Model, Master thesis, Institute of Biomedical Engineering, National Cheng Kung University, 1995.
[13] K. S. Arun, T. S. Huang, and S. D. Blostein, “Least-square fitting of two 3-D point sets,” IEEE Trans. Patt. Ana. Mach. Intell., vol.9, no.5, pp.698-700, 1987.
[14] S. C. Chapra and R. P. Canale, Numerical Methods for Engineers with Personal Computer Applications, McGraw-Hill, 1985.
[15] S. Theodoridis and K. Koutroumbas, Pattern Recognition, 2nded, Academic Press, 2003.