以人體視覺分類牙弓形狀(尖形，橢圓形，方型) 之統計分析

Objective: Many researchers have attempted to define ideal dental arch forms utilizing complex mathematical equations like parabolas, hyperbolas, and sixth power polynomials.The purpose of this study was to find a simple way to differentiate tapered, ovoid, and square arch forms. Material and Methods: Seventy-nine mandibular arch casts were obtained from students at Kaohsiung Medical University. Every tooth in each cast was computer digitalized into x and y coordinates. One orthodontist, one prosthodontist and eight post graduate students were invited to classify these 79 casts according to arch types (tapered, ovoid and square) based on their clinical experience. Ten distance variables ( 5 arch width and 5 arch depth variables) and fourteen ratio variables were tested. The 5 arch width variables included the distance between cusp tips of bilateral canines (W3), between buccal cusp tips of bilateral first premolars (W4), between buccal cusp tips of bilateral second premolars (W5), between mesio-buccal cusp tips of bilateral first molars (W6) and between disto-buccal cusp tips of bilateral second molars (W7). The 5 distance variables of arch depth included the distances from the central incisors’ incisal edge to the line joining the cusp tips of bilateral canines (D3), to the line joining the buccal cusp tips of bilateral first premolars (D4), to the line joining the buccal cusp tips of second premolars (D5), to the line joining the mesio-buccal cusps of the first molars (D6), and to the line joining the disto-buccal cusps of the second molars (D7). The 14 ratio variables included : W3/W7, W3/W6, W3/W5, W3/W4, W4/W7, W4/W6, W4/W5, (W3/D3)/(W7/D7), (W3/D3)/(W6/D6), (W3/D3)/(W5/D5), (W3/D3)/(W4/D4), (W4/D4)/(W7/D7), (W4/D4)/(W6/D6) and (W4/D4)/(W5/D5). The orthodontist’s classification was utilized as the basis for the correlation of variables. ANOVA was used to statistically establish significance between the 3 arch forms and the distance and ratio variables. Ratio variables that were statistically significant were selected for Receiver operating characteristic (ROC) curves analysis. Receiver operating characteristic (ROC) curves were computed using the SPSS statistical program to calculate the area under curve (AUC) for each chosen ratio variable. Larger AUC indicated better classification ability. AUC >0.8 indicate good classifying ablility. We used kappa statistic to calculate the strength of inter-observer agreement of arch form classification between the orthodontist and the other dentists. (kappa value : 0.75 : excellent) Result: The distance variables among males were longer than females, except D3. The results showed that there was no significant gender difference in ratio variables. However, there were significant differences among 6 distance variables (W3, W4, W5, W7, D3, D4) with arch form classifications when analyzed with ANOVA (p 0.4) . Conclusion: The ratio variable (W3/D3)/ (W7/D7) provided the best results in determining arch form. The Mean ± SD was 4.60± 1.15. Keywords: arch form, distance variables, ratio variables, Receiver operating characteristic (ROC) curves