Quantitative description of temporomandibular joint sounds: defining clicking, popping, egg shell crackling and footsteps on gravel.

Abstract

This study presents a quantitative description of temporomandibular joint (TMJ) sounds provided by a rule-based classification system based on sound classification by three dentists, who listened to and classified the sound recordings as no sound, click, coarse crepitus and fine crepitus. The sounds were recorded with microphones in the ear canal from 126 subjects during vertical opening, digitized at 15 000 Hz, and replayed using a computer sound card and speakers. The dentists' classification of a test set resulted in intra- and inter-tester j values ranging from 0.71 to 0.81 and 0.61-0.73, respectively. Pooled j values for the dentists and the dentists plus the rules were 0.67 and 0.58, respectively, which were not significantly different in terms of the sound features on which the rules were based (P = 0.13). Linear discriminant analysis showed the four TMJ sound types were significantly different (P < 0.001). The performance of the rules was equivalent to the dentists and marginally better than the linear discriminant functions (P = 0.08), establishing the validity of the quantitative descriptions they provide. The recording and rebroadcast methodology produced sounds very similar to those observed in the clinic and could be used to train clinicians in classifying TMJ sounds.