The influence of observer calibration in temporomandibular joint magnetic resonance imaging diagnosis

Objective. This study was undertaken to investigate the potential of reducing observer variation through a calibration program. Study design. The study was based on three sets of randomly selected temporomandibular joint magnetic resonance images. Each set consisted of bilateral images from 20 consecutive patients with temporomandibular disorders. As a baseline, three well-experienced noncalibrated investigators interpreted the images individually for disk position and disk configuration. After the initial interpretation, interobserver agreement was calculated as a kappa index and presented to the examiners. On the same occasion, the investigators analyzed agreement between them on the criteria to be used. Results. Overall data in this study showed an increase in the frequency of interobserver agreement with regard to disk position after the calibration trials were instituted. With regard to disk configuration, substantial interindividual variations were observed even after the observers reached consensus as to the criteria to be used. Conclusions. These data suggest that after calibration trials, it is possible for three examiners to obtain reliable and reproducible results in reporting temporomandibular joint disk position on magnetic resonance images.