Blind Separation of TMJ Sounds by Incorporating Fractional Delay Estimation

Temporomandibular joint (TMJ) sounds are generated from the two joints holding the lower jaw and the temporal bone (the skull) in patients suffering from temporomandibular disorder (TMD). In this study, we shall show how the inherent fractional delay between a pair of TMJ sources can be estimated by time localised sparse component analysis. In effect, the active periods of one of the sources will assist us in estimating the delay by employing modified versions of either the maximum likelihood (MLE) delay estimator or the mixed modulated Lagrange explicit time delay estimator (MMLETDE). Thereafter, we consider a strategy based on delaying the mixtures to solve the blind source separation of the TMJ sources. Simulation studies confirm the efficacy of the proposed scheme.