Electromyographic evaluation of a target muscle can be confounded by the mixture of signals from surrounding muscles, which interferes with data interpretation and degrades the reliability of assessment. In the present study, we applied independent component analysis (ICA), a method of multivariate analysis, to laryngeal electromyography in order to separate the muscle activity and enhance the quality of examination.
Laryngeal electromyography was recorded from a 37-year-old healthy male subject. Double-hooked bipolar electrodes were percutaneously inserted into the cricothyroid muscle (CT), sternohyoid muscle (SH) and their midpoint (CT+SH) as a contaminated signal. Electromyographic data during various tasks (falsetto voice /i/, glissando /i/, jaw opening and closing, neck flexion, falsetto /aiai/ and expiration with falsetto gesture) were recorded. Then, ICA was applied to the observed signals of CT and CT+SH to separate the original source signals of CT and SH. Finally, the separated SH signals were compared with the observed signals of SH.
ICA application improved the quality of the observed SH signals by 69.5%. Separated source signals generally coincided with the activation patterns of CT and SH reported in the literature. With a coefficient of determination of 65.1% between separated and original SH signals, the separation of source signals was performed with high precision.
The separation of source signals using fast ICA was feasible in laryngeal electromyography. ICA can be an effective analysis method to minimize undesirable signal mixture and enhance the quality of laryngeal electromyography.
View full abstract