Detecting functional coupling between electromyogram and local field potentials in dystonia using discrete wavelet transform and coherence estimation

In dystonia rhythmic activity and sustained hypertonic activity are often superimposed in the electromyograms (EMG) of affected muscles. We have used discrete wavelet transform (DWT) to decompose the EMG so that co-contraction between the agonist and antagonist muscles and functional coupling between EMG and local field potentials (FPs) in the internal globus pallidus could be detected. Using multiresolution analysis, the envelope frequency of EMG bursts was better determined by filtering between 50-200 Hz followed by rectification. This enhanced our ability to detect coherence between bursts in the EMG and local FPs recorded from the internal globus pallidus. Thus multiresolution analysis based on DWT is effective for detecting rhythmic EMG bursts in the time domain; and this increases signal-noise ratio for estimating coherence.