CLASSIFICATION OF TEMPOROMANDIBULAR-JOINT SOUNDS BASED UPON THEIR REDUCED INTERFERENCE DISTRIBUTION

Temporomandibular joint (TMJ) sounds were recorded in 98 orthodontic r etention patients, mean age 19 +/- 8.6 (s.d.) years, by interview, aus cultation and electronic recording. Sounds were found by auscultation in 41% and by interview in 32% of the subjects, more often in females than in males (P < 0.05). A new method for time-frequency analysis, th e reduced interference distribution (RID), was used to classify the el ectronic sound recordings into five subclasses, RID types 1-5, based u pon location and number of their energy peaks. RID types 1-3 had a few energy peaks close in time. RID types 4-5, typical of subjects with c repitation, had multiple energy peaks occurring close in time for a pe riod of 20-300 ms. RID type 1, found in 45% of the subjects, typical o f patients with clicking, had its dominant energy peak located in a fr equency range <600 Hz and was significantly more common in the female than in the male subjects (P < 0.01). RID type 2, found in 68% of the subjects, with the dominant peak in the range 600-1200 Hz, and RID typ e 3, found in 38% of the subjects, with the peak in the frequency rang e >1200 Hz, were found to have a similar gender distribution. RID type 4, found in 49% of the subjects, had the energy peaks distributed in the frequency range <600 Hz. RID type 5, found in 43% of the subjects, more often in females than in males (P < 0.05), had the peaks distrib uted over the whole frequency range from about 30 Hz up to about 3000 Hz. In conclusion, a more detailed classification could be made of the TMJ sounds by displaying the RIDs than by auscultation. This suggests that RID classification methods may provide a means for differentiati ng sounds indicating different types of pathology.