Automated analysis of electromyographic (EMG) recordings during botulinum injections

Botulinum toxin is injected into extraocular muscles using electromyographic (EMG) guidance to confirm needle location prior to injection. The purpose of this study was to develop and test an automated, objective method of storing and grading EMG signal quality obtained during botulinum injection. A reliable, nonsubjective estimate of signal quality will enhance future studies of the complications or effectiveness of botulinum toxin injections into extraocular muscle.

Injections were administered using a standard procedure. The EMG was digitized, recorded, and analyzed using discrete Fourier transform and power spectrum analysis. Audible signal quality was also graded subjectively by the surgeon. Patients were followed to determine the clinical response to the injection.

Eighteen patients (24 injections) were studied. Each sample was collected and analyzed in less than one second. The EMG contained signals over a range of frequencies from 50 to 500 Hz, with an average peak near 90 Hz. Substantial variability of the signal was observed between patients. Contracting muscle had a higher total power and peak frequency in most, but not all, cases. Total signal power correlated moderately with subjective estimates of signal quality. Neither subjective nor computer-derived signal quality estimates correlated with the response to injection or complications.

Computerized signal analysis can automatically characterize EMG signal power and frequency distribution. Although this technique has the potential to identify needle location at the time of botulinum injection accurately, signal variability limited the clinical and research utility of the present protocol.