Using Artificial Intelligence to Evaluate the Impact of Glabellar Paralysis on Facial Expression of Emotion
Many patients are concerned about emotional expression after onabotulinum toxin A injection. We sought to investigate the effect of injection on expression using artificial intelligence.
Methods/Technique
:Fifty-two female subjects (“expressors”) were recorded while viewing video clips intended as a mood induction procedure that stimulates surprise and anger emotions. The subjects were then injected with 25 units of onabotulinum toxin A in the glabellar region. All video clips from time periods before injection and 1 month after injection from the 10 maximal expressors were extracted and shown to a group of 31 “perceivers” who rated the facial expressions for intensity (Likert 1-7) and identity of emotion (percent correct emotion identified). The video expressions detected within each clip were analyzed using a commercially available facial expression recognition software package (FaceReader™, Noldus Information Technology BV, Wageningen, Netherlands). The data generated from the software reflected the proportion of each emotion expressed for any given facial movement and the action units associated.
Results/Complications
:Glabellar paralysis significantly diminished mean perceived intensity of anger (50.4% relative reduction, p<0.001) and surprise (20.6% relative reduction, p<0.001). For anger videos before and after botox injections, angry emotions detected by the software decreased from 13.71% to 1.47% (p=0.062), and surprise emotions decreased from 5.08% to 4.2% (p=0.75). For surprise videos before and after botox injections, angry emotions detected decreased from 7.67% to 1.84% (p=0.75), and surprise emotions increased from 5.95% to 7.65% (p=0.125).
Conclusion
:
This study provides the first proof of concept for the use of a machine learning software application to objectively quantify facial expression before and after botox injections. The software results are consistent with human raters in detecting changes in anger, but give the opposite effect in term of surprise.