臺灣博碩士論文加值系統

牙齒敏感必須從牙齦線是否有萎縮與琺瑯質厚度去判斷。由於各醫院每天所產生的X光片影像數量非常的可觀，所以需要電腦輔助來幫助醫生診斷。然而X光片影像可能因人為操作不當，使得影像品質不佳，故需要利用影像處理方法才能將牙齦線位置與琺瑯質厚度正確取出。
本論文在牙齦線偵測法中提出的方法是先以二值化的方式將牙齦線的大略範圍圈選為視窗(window)，再以Canny邊緣運算取得window內可能的線段，藉由線段的位置，長短，上下方區域計算出平均強度及紋理(對比度)篩選出適合的牙齦線，最後再利用左右兩測的位置判定是否取出正確的牙齦線。
琺瑯質量測上藉由SMQT影像增強方法將琺瑯質區域顯示，琺瑯質厚度量測則是藉由Canny 邊緣偵測將琺瑯質的邊緣找出，再分別計算深度與厚度。本論文在牙齦線偵測上總共測試十二張原始X光片影像，將其分割成單顆牙齒共有三十張，實驗結果顯示，本論文所提的方法與人眼仔細判定的結果比較，有8張完全吻合，4張有1或2顆不完全吻合。計算琺瑯質深度與厚度的X光片影像則有12張，有7張完全吻合，3張不完全吻合，2張完全不吻合。

關鍵字： 牙齦線偵測、琺瑯質流失偵測、Canny邊緣運算

Sensitive tooth or periodontal disease both can be detected through early detection of gum-line receding and enamel loss. Because vast amount of dental radiographs being produced each day, and quality of X-ray images may not always be good due to improper operations when taking the radiograph. Thus, doctors often need computer aided system for better diagnosis. In this paper, we propose a method of gum-line extracting and a method of enamel measurement.
For gum-line detection, we first apply thresholding to roughly separate the background from tooth and gum, and divide the binary image into left and right halves. Using the area between the gum point and crown in each half image as the region of interest (ROI), we apply Canny edge filter to retrieve all possible edge lines. We then filter these line candidates by their angle, length, position, and texture of their neighborhood area and obtain a final gum line. For better result, we verify the y-position of both left and right gum lines.
For enamel measurement, we adopt an image enhancement method SMQT to stretch the contrast between enamel and dentin then apply Canny edge operator to isolate enamel from dentin. Finally, the depth and the thickness of enamel are measured by taking the maximum height and the maximum width of the enamel, respectively. Among all 12 dental radiographs tested for gum-line detection, eight images have results match completely with human perception, four images have results match partially, and none fails completely. Among all 12 dental radiographs tested for enamel measurement, seven images have results match completely with human perception, three images have results match partially, and two fail completely.
Keyword: Gum-line detection, Enamel-loss detection, Canny edge detection

摘 要 i
ABSTRACT ii
誌 謝 iii
目 錄 iv
表 目 錄 v
圖 目 錄 vi
第一章、緒論 1
1.1研究動機與目的 1
1.2研究背景 1
第二章、相關文獻 4
第三章、牙齦線偵測 9
牙齦線判定 9
3.1牙齒分隔 10
3.2背景分離 11
3.3牙冠到牙齦之間的視窗(window) 13
3.4線段篩選 16
3.5確認牙齦線位置 20
牙齦線流程圖 21
第四章、琺瑯質偵測方法 22
4.1 DEJ橫向邊緣偵測 23
4.2 DEJ縱向邊緣偵測 25
4.3計算琺瑯質寬度與厚度 26
第五章、實驗結果與分析 27
5.1牙齦線偵測實驗結果 27
5.2琺瑯質偵測實驗結果 30
第六章、結論與未來方向 33
參考文獻 34

[1] http://www.tda.org.tw/index.php?option=com_content&view=article&id=516:2008
-11-05-12-43-34&catid=107:pain&Itemid=109
[2]Dr. Tony Information Technology Corporation, http://www.dr-tony.com.tw/
[3]J. Zhou and M. Abdel-Mottaleb, “A content-based system for human identification based on bitewing dental X-ray images,” Pattern Recognition Vol. 38, pp. 2132-2142, 2005.
[4]N. Otsu, “A threshold selection method from gray-level histogram,” IEEE Transactions on Systems, Man, and Cybernetics. pp. 62-66, 1979.
[5]J.F Canny, “A Computational Approach to Edge Detection,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol 8, pp. 679-698, 1986.
[6]M. Nilsson, F. Sattart, H. K. Chngt, “Automatic Enhancement and Subjective Evaluation of Dental X-ray Images using the SMQT,” IEEE Information, Communications and Signal Processing, pp. 1448-1451, 2005.
[7]J. Hua, S.K Chen, and Y. Kim, “Refining Enamel Thickness Measurements from B-Mode Ultrasound Images,” IEEE Engineering in Medicine and Biology Society, pp. 440-443, 2009.
[8] O. Nomir and M. Abdel-Mottaleb, “A System for Human Identification from X-ray
Dental Radiographs,” Pattern Recognition, Vol. 38, pp. 1295-1305, 2005.
[9] Gonzalez,Wood著，謬紹綱譯，數位影像處理 3e, pp. 657-659