Analysis of biological signals by Gabor decomposition

Joint time-frequency representations of data from two unrelated experiments have been providing clinicians important insight into signal characteristics which distinguish one signal category from another. Though the source of the signal in the two experiments is quite different, common signal features have been identified which motivated the analysis method presented in this paper. In the first experiment, event-related potentials (ERPs) were recorded from psychiatric patients. Temporomandibular joint (TMJ) sounds were recorded from dental patients in the second experiment. A signal feature common to both sets of data is one or more energy concentrations localized in time and frequency. It is this feature which is captured by the Gabor decomposition method described in this paper