陳羿貞Chen, Yi-Jane臺灣大學:臨床牙醫學研究所黃榆鈞Huang, Yu-ChunYu-ChunHuang2010-05-262018-07-092010-05-262018-07-092009U0001-1808200910125800http://ntur.lib.ntu.edu.tw//handle/246246/184177目的：　顱顏面不對稱是一項咬合異常患者常見的特徵，嚴重的顏面不對稱可能需要正顎手術合併矯正治療才能予以改正。為了診斷不對稱發生的區域，傳統上，我們是用二維影像系統作為診斷工具，如側面測顱片或是前後向測顱片。由於最近錐狀射束斷層掃描影像（CBCT）的發展，我們可以更便利的得到三維立體影像，進而得到更多的診斷資訊。在本研究中，我們希望能在釐清CBCT系統中界標辨識的可重複性及測量的正確性，並試圖以新的分析方法來研究顱顏面不對稱的情況，並提出適切的分類方式。驗材料及方法：研究的第一部份主要是希望探討錐狀射束電腦斷層掃描和數位測顱片的準確度。首先利用電子游標尺在三個乾燥顱骨上測量17個線性測量值，同時將顱骨以三種不同的方式得到影像，分別為測顱片手繪測量、電腦軟體測量（Winceph 8.0）及三維影像測量（i-CAT/ Simplant Pro 12.02）。以真實測量值為黃金標準，計算各種方法的組內相關性，以釐清其測量值的可信度及影像之放大率。二部分則是分析臨床矯正患者的錐狀射束電腦斷層掃描三維影像，以評估顏面不對稱的情形。我們共收集33位患者之三維影像資料並分為兩組，Group 1為顏面較為對稱的患者，在正面測顱片上，其Gn與矢狀切線的距離小於4mm，而Group 2中的患者則是有較明顯的不對稱，且Gn的偏移量大於4mm。在三維影像的處理上，先定義正中矢狀切面的位置為經過ANS、Cg及Ba點之平面，而後在軟體(i-CAT/ Simplant Pro 12.02)中辨識出所選定的11個界標點，並記錄其座標；在三個不同的方向上，分別計算左右兩側座標值的差異；此外，我們也利用問卷的收集，評估不同專科經驗的醫師對顱顏面不對稱的判別情形。所採用的統計方法包括了二元迴歸分析及群聚分析法PAM群聚分析法（partition around medoids cluster analysis），進一步釐清顏面不對稱的各種不同表現情形。果：在本研究中，比較真實值與三種不同影像來源所得到的線性測量結果，發現彼此都存在著極高的相關性，且達統計上的顯著性，其中又以CBCT所得到的相關係數最高，再者為軟體測量，最後則為手繪測量。於第二部份的研究，由迴歸分析所得到的結果，我們發現所選出的不對稱預測因子，包括有：△Hco、△Area 1、△Area 2、△APJr及△HCo-Go。而用群聚分析對患者不對稱的特徵進行分群，則發現可以分為五群。其中Cluster(5)-1是其中最為對稱的一組，而Cluster(5)-3、5則最為不對稱，但兩組擁有不同的特徵。Cluster(5)-2及4的不對稱情形則居中，其中Cluster(5)-2沒有明顯的代償現象，而在Cluster(5)-4則主要的不對稱因子是來自於下顎骨體的差異，但在其他項目上則有明顯的代償現象。而從問卷的結果得知，相較於牙科部實習醫師及齒顎矯正專科醫師，在矯正科受訓中的住院醫師傾向於更容易辨別出顏面不對稱，也可能以較嚴格標準進行顏面不對稱的判定，且更傾向於建議病人接受正顎手術的治療，此外，我們應用了GAM（generalized additive model）技術來協助鑑定不對稱的主觀判定閥值，發現本研究所採用的Gn偏移量4mm為一個適切的標準。而最終的迴歸模式分析結果，則得出建議接受正顎手術的相關預測因子包括有：下顎偏移量、醫師的經驗、群聚分析分類及上顎是否偏斜。論：真實值為基準，不論是用軟體測量（Winceph）或是CBCT的直接測量，我們都可以證實其間的高度相關性。以錐狀射束電腦斷層掃描三維影像評估齒顎矯正患者顏面不對稱的結果可知，本研究所提出的群聚分析法，是一個合適的分類法，有助於我們正確診斷各種表現不同的顏面不對稱並做出合適的治療計畫。文關鍵詞：錐狀射束電腦斷層掃描、顱顏面不對稱、測顱分析、三維影像、群集分析Objectives: Craniofacial asymmetry is a common feature in human. Orthodontic treatment combined with orthognathic surgery would be suggested in some extreme cases. In the past, 2D images, such as posterioanterior cephalogram or panoramic film, were used as the diagnostic tools to determinate the asymmetric component in craniofacial area. Since the cone beam CT (CBCT) was developed, 3D image would be available in our daily practice. The objectives of this research were to evaluate the reproducibility of landmark identification and the reliability of measurement in CBCT. Next, we developed a screening strategy for analyzing different types of facial asymmetry. ethod: Part I, 17 linear dimensions of 3 dry skulls measured by digital caliper were used as gold standards. Hard copies of manual tracing, computer-aided digital measurements (Winceph 8.0) of both lateral and PA cephalograms, and 3D direct measurements in CBCT (i-CAT/ Simplant Pro 12.02) images were compared with the gold standards. Interclass correlation coefficient was used to evaluate the reliability and magnification in different modalities. Part II. The CBCT data of 33 subjects was collected in this study and classified into 2 groups. The subjects in group 1 showed acceptable level of asymmetry and the amount of Gn deviation to midsagittal line was less than 4mm in frontal cephalogram. Those in group 2 showed obvious asymmetry with Gn deviation more than 4mm. In 3D dataset, the midsagittal plane was determinated as the plane passing ANS, Cg and Ba. Other 11 landmarks were identified and their coordinates were recorded. The differences of the distances between landmarks to reference plane bilaterally were calculated in 3 dimensions. Questionnaire was used to evaluate the subjective opinion of orthodontist at different training levels. Binary regression analysis and cluster analysis (PAM) were used for statistical analyses. esult: Part I, three modalities in our study all showed statistically significant correlation with gold standard. However, the highest one was CBCT, and manual tracing was the least. Part II, binary regression model showed that the predictors for grouping of facial asymmetry included △HCo, △Area 1, △Area 2, APJr and △HCo-Go. In cluster analysis, all subjects were classified into 5 clusters. The most symmetric cases were grouped into cluster (5)-1. The cluster(5)- 3 and cluster(5)- 5 were obvious asymmetric cases with different characteristics. The severity of asymmetry was moderate in cluster(5)-2 without obvious compensated components. In cluster(5)- 4, the major asymmetric component was the difference of area 2, but other variables compensated the manifestation of asymmetry. In the result of questionnaire, the residents in orthodontic department showed higher tendency than specialists and interns in identifying facial asymmetry and suggesting patients to receive orthognathic surgery. On the other hand, GAM proved the 4mm deviation of Gn as a proper threshold to identify the facial asymmetry for its adequacy. Moreover, according to the binary regression model, the factors responsible for the suggestion of orthognathic surgery included the amount of chin deviation, training level of doctors, clustering category and the presence of maxillary canting.onclusion: The reliability of measurement in Winceph system and in CBCT had high correlation with virtual value. The cluster analysis could be applied to classify the patients with facial asymmetry for diagnosis and treatment planning.ey words: cone bean computed tomography, craniofacial asymmetry, cephalometric analysis, 3D image, cluster analysis口試委員會審定書 i謝 ii文摘要 iiibstract v一章 緒論 1.1 研究背景 1.2 顱顏發育之不對稱 2.2.1顱顏發育不對稱之綜論 2.2.2 顱顏面不對稱之診斷 2、外觀影像 3、側面測顱片 3、環口X光攝影 4、正面測顱片 4、Submentovertex測顱片 5、使用三維影像 6.3 錐狀射束電腦斷層掃描儀 7.3.1 現今市面上可見之錐狀射束斷層掃描儀 7.3.2 錐狀射束電腦斷層掃描技術的發展對於矯正學的影響 8、輻射劑量的考量 8、臨床應用與近期研究 9.4醫學影像之準確性探討 10.4.1測顱分析在齒顎矯正診斷上之應用 10.4.2 錐狀射束電腦斷層掃描於臨床應用上之準確性 14.5研究目的 15二章 研究材料與方法 16.1錐狀射束電腦斷層掃描、數位測顱片之手繪測量與電腦軟體測量的線性測量正確性之比較 16.1.1研究樣本 16.1.2研究樣本置備 16.1.3 測量方式及工具 18.2利用三維影像分析顱顏面不對稱之情形 21.2.1研究對象 21.2.2研究材料製備 21.2.2.1正面測顱影像 21.2.2.2 分組 21.2.2.3 三維影像資料之取得 22.2.2.4 觀察者誤差（Intraobserver error） 24.2.2.5外觀照片分析及問卷調查統計 25.2.2.6統計分析 25三章 結果 28.1 錐狀射束電腦斷層掃描、數位測顱片之手繪測量與電腦軟體（Winceph）測量於線性測量上之比較 28.1.1 線性測量值正確性之比較 28.1.2 影像放大率 29.2利用三維影像分析顱顏面不對稱之情形 30.2.1 界標判定之觀察者組內誤差 30.2.2 顱顏面不對稱之迴歸分析模型 30.2.3 群集分析 31.2.4 問卷結果分析 32.2.4.1 卡方分析 32.2.4.2不同經驗醫師對是否需接受正顎手術治療判定結果之迴歸分析模型 32四章 討論 34.1錐狀射束電腦斷層掃描測顱分析、數位測顱片之手繪測量與電腦軟體測量之比較 34.1.1線性測量值正確性之比較 34.1.2影像放大率之探討 35.2利用三維影像分析顱顏面不對稱之情形 36.2.1 正中矢狀平面之定義 37.2.2 界標判定之觀察者組內誤差 39.2.3 顱顏面不對稱之迴歸分析模型 42.2.4 群集分析 44.2.5 問卷結果分析 46五章 結論 49六章 未來展望 50考文獻 86application/pdf2610038 bytesapplication/pdfen-US錐狀射束電腦斷層掃描顱顏面不對稱測顱分析三維影像群集分析cone bean computed tomographycraniofacial asymmetrycephalometric analysis3D imagecluster analysishttp://ntur.lib.ntu.edu.tw/bitstream/246246/184177/1/ntu-98-R95422023-1.pdf