Development of a new category system for the profile morphology of temporomandibular disorders patients based on cephalograms using cluster analysis

OBJECTIVE: This study aims to develop a new category scheme for the profile morphology of temporomandibular disorders (TMDs) based on lateral cephalometric morphology. METHODS: Five hundred and one adult patients (91 males and 410 females) with TMD were enrolled in this study. Cluster tendency analysis, principal component analysis and cluster analysis were performed using 36 lateral cephalometric measurements. Classification and regression tree (CART) algorithm was used to construct a binary decision tree based on the clustering results. RESULTS: Twelve principal components were discovered in the TMD patients and were responsible for 91.2% of the variability. Cluster tendency of cephalometric data from TMD patients were confirmed and three subgroups were revealed by cluster analysis: (a) cluster 1: skeletal class I malocclusion; (b) cluster 2: skeletal class I malocclusion with increased facial height; (c) cluster 3: skeletal class II malocclusion with clockwise rotation of the mandible. Besides, CART model was built and the eight key morphological indicators from the decision tree model were convenient for clinical application, with the prediction accuracy up to 85.4%. CONCLUSION: Our study proposed a novel category system for the profile morphology of TMDs with three subgroups according to the cephalometric morphology, which may supplement the morphological understanding of TMD and benefit the management of the categorical treatment of TMD.