Noncontact Sensing of Facial Muscle Activity Using Laser Doppler Vibrometry: Time Domain Data Analysis

Abstract

Movements of the facial muscles offer a promising avenue for assessing stress, fatigue and emotion, and appear useful for a number of applied and clinical purposes. This paper describes a novel application of laser Doppler vibrometry (LDV) as a noncontact method for assessing facial myographic activity. The principle of the LDV method involves detection of the minute vibrations of contracting muscles, associated with the activation of individual motor units. Data were obtained from 11 participants who received 15-20 min of training before the measurement acquisitions. Participants produced several standardized facial expressions involving activation of the upper face (lowering and raising the eyebrows) and the lower face (raising the upper lip, stretching the lip corners, and clenching the jaw). The associated muscle vibratory activity was assessed using the LDV method, within context of the simultaneous electromyogram (EMG). A separate condition entailed study of the jaw muscle signals during repetitive chewing. The temporal, spatial and measurement sensitivity aspects were studied in separate tests; the present report focuses on the temporal aspects of the response, in comparison to the onsets and offsets of the simultaneous EMG and gross facial surface displacement. The LDV signals were obtained from a site overlying the principal involved muscle for the various movements. Results showed that LDV myographic signals (LDV-MMG) could be recorded from all facial muscles studied, although they were relatively small from the muscles of the upper face. LDV-MMG signals were especially prominent at times of contraction onset and offset, indicating that the method may be particularly useful for the study of dynamic activity as would be associated with brief changes in facial expression. The LDV-MMG signals generally were found to lead the onset of the EMG signal by about 100 ms, and to lag the offset of the EMG signal by about 200 ms. The LDV-MMG response associated with chewing was associated in time with the EMG and displacement signs of chewing, but were generally more polyphasic in form. The findings generally support the potential use of the LDV method as a non-contact and non-obtrusive method for assessing activity of the facial muscles.