Comparative study on ordinary coordinates and standardized coordinates converted by general procrusts analyzation method in cluster analysis and discrimination analysis of AngleⅠ malocclusion / 北京大学学报(医学版)

ABSTRACT

Objective: To classify the AngleⅠ malocclusion subgroups using lateral radiographic films according to their ordinary coordinates and standardized coordinates converted by general procrusts analysis (GPA), and to compare the two kinds of classifications and their valne in diagnosis of malocclusion. Methods: 946 pretreatment lateral radiographic films of patients with Angle'I malocclusion were chosen and their ordinary coordinates were acquired via soft ware. The ordinary coordinates were then converted into standardized coordinates by GPA. All of the films were classified by cluster analysis and discrimination analysis applying the upper two kinds of coordinates respectively. Results: (1) Twenty one subgroups were identified according to the ordinary coordinates of the chosen films with the total differentiate rate of 92.7% and leave-one-out differentiate rate of 68.4% (Classification A). Correspondingly, 20 subgroups were identified according to the standardized coordinates with the two differentiate rates of 87.8% and 71.9% (Classification B). (2) If the ordinary coordinates were discriminated by Classification B, the total differentiate rate and leave-one-out differentiate rate were 79.8% and 60.2 % respectively. If the standardized coordinates were discriminated by Classification A, the two differentiate rates were 79.8% and 60.2 % respectively. (3) There were some subgroups having the similar form in Classification A and their difference mainly arose from the difference of the patient age, while there were no such subgroups like that in Classification B. (4)The proportion of the largest subgroup in total subjects is 15.9% and there were 8 subgroups having the number of subjects over 40 in Classification A, whilethe corresponding proportion and number of subgroups were 74.7% and 2 in Classification B.(5)Classification A and Classification B were both suitable to classify a new subject, but Classification B was required to standalize all of the subjects onc again, which was very complex, while Classification A was more simplified. Conclusion: Classification A and Classification B are interelated. GPA could concentrate the subjects in cluster analysis, which reduces some influence of the age to the classification, but the process to classify a new subject is very complex. Thus if quick diagnosis is needed in clinics, Classification A should be recommended, but the influence of the age should be noted.