Mental workload (MW) represents the brain resources an individual devotes to a task. The evaluation of MW is fundamental for Advanced driver-assistance systems (ADAS). To assess MW, non-invasive techniques are preferable to avoid interference with the driver. Infrared Thermal Imaging (fIRI) is highly suited given its contactless nature. The research reported aimed at developing a contactless physiological method to measure MW employing fIRI features obtained from human facial skin temperature modulations. The novelty of this study is that MW was evaluated employing functional Near Infrared Spectroscopy (fNIRS), a non-invasive optical technique that measures the hemodynamic oscillations related to cortical activations. Particularly, the Sample Entropy of the fNIRS signal was assumed as indicative of MW. A two-level (i.e. High MW vs. Low MW) Support Vector Machine (SVM) classifier with linear kernel was employed to predict the level of MW from fIRI features. A leave-one-subject-out cross-validation was implemented to test the generalization performances of the method. The classifier showed a cross-validated sensitivity of 77% and specificity of 69%. This study represents the first attempt to estimate MW evaluated by fNIRS from fIRI features.
Can Functional Infrared Thermal Imaging Estimate Mental Workload in Drivers as Evaluated by Sample Entropy of the fNIRS Signal?
Perpetuini D.;
2021-01-01
Abstract
Mental workload (MW) represents the brain resources an individual devotes to a task. The evaluation of MW is fundamental for Advanced driver-assistance systems (ADAS). To assess MW, non-invasive techniques are preferable to avoid interference with the driver. Infrared Thermal Imaging (fIRI) is highly suited given its contactless nature. The research reported aimed at developing a contactless physiological method to measure MW employing fIRI features obtained from human facial skin temperature modulations. The novelty of this study is that MW was evaluated employing functional Near Infrared Spectroscopy (fNIRS), a non-invasive optical technique that measures the hemodynamic oscillations related to cortical activations. Particularly, the Sample Entropy of the fNIRS signal was assumed as indicative of MW. A two-level (i.e. High MW vs. Low MW) Support Vector Machine (SVM) classifier with linear kernel was employed to predict the level of MW from fIRI features. A leave-one-subject-out cross-validation was implemented to test the generalization performances of the method. The classifier showed a cross-validated sensitivity of 77% and specificity of 69%. This study represents the first attempt to estimate MW evaluated by fNIRS from fIRI features.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
