Researchers Use AI and MRI to Create Facial Expression Pain Signature
Facial expressions play an important role in communicating the unpleasant sensory and emotional experience of pain. Among other things, they signal to others that we are hurt and may need help.
The neural processes associated with this form of nonverbal expression have received little attention although they are known to play an important role in the experience of pain. Marie-Eve Picard, a doctoral student in the laboratory of Pierre Rainville, a professor in the Faculty of Dentistry at Université de Montréal and a researcher at the Montreal University Institute of Geriatrics Research Centre, decided to investigate.
In a new study, Picard and Rainville show that facial expressions triggered by painful stimuli can be predicted from brain activity. Their findings reveal that the neural mechanisms underlying these expressions are largely distinct from those associated with other manifestations of pain, such as subjective verbal reports of perceived intensity.
Picard and her colleagues developed a neurobiological model that predicts facial expressions elicited by painful stimuli. Using machine-learning algorithms trained on magnetic resonance brain imaging data, they created a Facial Expression Pain Signature.
Healthy volunteers underwent painful thermal stimulation and their facial expressions were measured using the Facial Action Coding System, a standardized tool that analyzes facial movements based on the activity of several groups of facial muscles.
Activation of each muscle group causes a specific change in facial expression. For example, pain-related expressions often include furrowed brows, elevated cheeks, squinting, wrinkled nose and raised upper lip.
In clinical settings, accurately assessing a patient’s pain is important for appropriate pain management.
“The importance of facial expression in pain assessment receives less attention than the role it plays in social interactions,” said Picard. “However, our results suggest that this behavioral indicator of pain can be a valuable complement to verbal reports of perceived intensity.”
The study was informed by an understanding of pain as multidimensional, meaning that considering its various manifestations can improve assessments of its severity.
Picard’s work shows the existence of brain signatures, or patterns of brain activity, that are predictive of pain-related facial responses. While these results advance our understanding of the brain mechanisms behind pain and nonverbal communication, further research will be needed to test their generalizability and determine their applicability to conditions such as chronic pain.