ThermICA: Novel Approach for a Multivariate Analysis of Facial Thermal Responses

Area of Science:

• Biomedical Signal Analysis
• Psychophysiology
• Medical Imaging

Background:

• Infrared Thermography (IRT) is a contactless method for monitoring skin temperature.
• IRT is utilized in clinical medicine and psychophysiology.
• Existing methods lack detailed autonomic correlate information.

Purpose of the Study:

• Introduce a new methodology using IRT to extract autonomic correlates.
• Obtain information on perspiration, peripheral vasomotility, and respiration.
• Enhance objective assessments of emotional states.

Main Methods:

• Model-based decomposition of facial thermograms.
• Utilized Independent Component Analysis (ICA) and ad-hoc preprocessing.
• Tested on 30 healthy volunteers under stimulated psychophysiological states.

Main Results:

• Identified three independent components correlating with physiological signals.
• Linear combination of components predicted electrodermal activity (r=0.9), respiration (r=0.8), and photoplethysmography (r=0.73).
• Cross-validated inter-subject analysis successfully predicted signals from unseen subjects' facial thermograms.

Conclusions:

• Validated efficacy of thermal-derived signals for differentiating experimental conditions.
• Demonstrated sensitivity and promise of the IRT approach for biomedical signal analysis.
• Highlighted potential for enhanced objective assessment of emotional states.