Classification of temporomandibular joint sounds based upon their reduced interference distribution

Abstract

Temporomandibular joint (TMJ) sounds were recorded in 98 orthodontic retention patients, mean age 19 ± 8–6 (s.d.) years, by interview, auscultation 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, the reduced interference distribution (RID), was used to classify the electronic sound recordings into five subclasses, RID types 1–5, based upon 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 crepitation, had multiple energy peaks occurring close in time for a period of 20–300 ms. RID type 1, found in 45% of the subjects, typical of patients with clicking, had its dominant energy peak located in a frequency 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 type 3, found in 38% of the subjects, with the peak in the frequency range >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 distributed 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 differentiating sounds indicating different types of pathology.