Previous authors have suggested that various physiological parameters may used to aid in the diagnosis of various temporomandibular joint pathologies. OBJECTIVE: To use a GMDH neural network to evaluate the contributions of 18 measured physiologic findings to the identification of specific TMJ pathologies. METHODS: Jaw movements and joint vibrations were recorded from forty-six TMD patients; 1) while chewing gum and 2) moving throughout their range of motion. Each patient's TMJs were verified as normal; anterior, antero-medial or medial disk displacement with reduction; disk displacement without reduction; early or late-stage degenerative joint disease. A GMDH neural network was trained repeatedly from the extracted dataset until each parameter was evaluated in relation to the five common pathological conditions and normal data. RESULTS: Important contributions (R squared values ranged from 0.76 to 0.96) were found from six to twelve of the parameters for each joint condition. Two parameters never contributed significantly to the process for any condition. While each of the other sixteen parameters contributed significantly to a strong correlation with at least one pathology, only the “maximum range of motion measurement” contributed significantly to all of the correlations. Six of the eighteen parameters (the smallest number) contributed significantly (R squared = 0.895) to identifying anterior disk displacement with reduction, whereas twelve (the largest number) contributed to the correlation (R squared = 0.922) with advanced degenerative joint disease. CONCLUSION: Since only one of the eighteen physiological parameters chosen contributed significantly to establishing a correlation with every condition tested, the use of physiological data to evaluate TM joints requires careful selection and validation of parameters.

Back to the Diagnostic Systems Program
Back to the 81st General Session of the International Association for Dental Research (June 25-28, 2003)

Top Level Search