Classification of Temporomandibular disorder from electromyography signals via Directed Transfer Function

A new approach for the detection of Temporomandibular joint disorders (TMDs) from the recorded electromyography (EMG) signals from the muscles around the temporomandibular joint (TMJ) has been presented in this paper. Multivariate Autoregressive (MVAR) modelling has been applied to a six-channel set of EMG signals from the muscles of both sides of the jaw during mouth opening and closing. The MVAR coefficients are then used to define the Directed Transfer Function (DTF), which estimates the strength of the direction of the signals flow between the channels. The DTF energy parameters were chosen as the features for EMG classification using support vector machine (SVM). The method described here has a potential to detect and classify the type and level of muscular disorder from the way the muscle signals interact with each other.