RESUMEN
INTRODUCTION: The purpose of this study was to present and validate a novel semiautomated method for 3-dimensional evaluation of the temporomandibular joint (TMJ) space and condylar and articular shapes using cone-beam computed tomographic data. METHODS: The protocol for 3-dimensional analysis with the Checkpoint software (Stratovan, Davis, Calif) was established by analyzing cone-beam computed tomographic images of 14 TMJs representing a range of TMJ shape variations. Upon establishment of the novel method, analysis of 5 TMJs was further repeated by several investigators to assess the reliability of the analysis. RESULTS: Principal components analysis identified 3 key components that characterized how the condylar head shape varied among the 14 TMJs. Principal component analysis allowed determination of the minimum number of landmarks or patch density to define the shape variability in this sample. Average errors of landmark placement ranged from 1.15% to 3.65%, and none of the 121 landmarks showed significant average errors equal to or greater than 5%. Thus, the mean intraobserver difference was small and within the clinically accepted margin of error. Interobserver error was not significantly greater than intraobserver error, indicating that this is a reliable methodology. CONCLUSIONS: This novel semiautomatic method is a reliable tool for the 3-dimensional analysis of the TMJ including both the form and the space between the articular eminence and the condylar head.