Comparison of ANB angle correction methods

The aim of this study is to obtain clusters which are pointing different craniofacial patterns, from the data pool that obtained from individuals with different sagittal orientation relations according to the ANB angle and to examine the differences of these clusters. A total of 330 lateral cephalograms were used in the study, there were 3 groups and each group contains 110 samples which have Class 1, Class 2, and Class 3 relations according to the ANB angle classification. 4 clusters were created by using the X-means algorithm of the Rapidminer program on all datas. To determine the relationship between the clusters and datas, decision tree analysis applied on Rapidminer. The same procedure was repeated on 288 lateral cephalograms after the extraction of individuals which are on the extreme branches of decision tree. The same procedure was repeated on the new data set and a new decision tree with 4 sets was created. The measurement that determines the first segregation point of the branch in the last decision tree was individual ANB calculation. When individual ANB values at the branches were larger than -2.3 , cluster 0 and cluster 1 has been found and when the values were smaller than -2.3, Cluster 2 and Cluster 3 has been found. The parameters that determine the difference between cluster 0 and cluster 1 were ; Interincisal angle, SNA °, SNB ° and the parameters that determine the difference between cluster 2 and cluster 3 were; N-Go-Gn °, SN / MP ° and SN / GoGn measurements. The differences between the clusters were analized by using ANOVA, Tukey HSD, Tamhane, Kruskal-Wallis, Mann Whitney U tests. There was a statistically significant difference in all measurements except N-Go-Ar °, FH / PP °, S-Ar-Go ° (p <0.05).