| Description |
Monitoring workload and fatigue using wearable sensors is very valuable in several conditions involved during everyday life, and even more in specific applications requiring attention during prolonged periods of time (e.g. during the surgical act, during a prolonged mission accomplished by an astronaut or a soldier). In this context, one can take advantage of light and portable devices, able to measure bioelectrical neuronal activity with Electro-EncephaloGraphy (EEG) together with cortical hemodynamic processes with Near InfraRed Spectroscopy NIRS. EEG measures on the scalp bioelectrical discharges elicited by synchronous neuronal populations, whereas NIRS monitors brain hemodynamic processes, exploiting the complementary absorption properties of oxy-hemoglobin (HbO) and deoxyhemoglobin (HbR), by sending and receiving infra-red light through head tissues from scalp sensors (at 690 and 830nm). As a bedside and portable brain-imaging device, NIRS is a promising clinical tool, complementary to functional Magnetic Resonance Imaging, fully adapted to monitor accurately brain activity in realistic lifestyle conditions, i.e. while walking, during physical exercise or gait perturbation, or while sleeping.
The proposed project aims at assessing the ability of EEG and NIRS to monitor attention and workload in fully controlled conditions first (e.g. memory, attention, mental calculation task), and then while performing a "Multi-attribute task battery (MATB)", a multi-tasking navigation simulator program developed by the NASA that requires the operator to simultaneously monitor and respond to several tasks (Targeting, Systems Monitoring, Communications, and Resource Management). MATB is a specific task mimicking situation that would occur when assessing workload during the mission of a soldier or an astronaut. A significant amount of data has been acquired on healthy participants at Concordia PERFORM centre (data collection still in progress). The objective of a first team will be to carefully preprocess, denoise (filtering, motion correction) and analyze EEG/NIRS data to extract typical brain responses assessing changes in cognitive workload. The second team will extract specific features relevant to train an algorithm able to detect specific levels of cognitive workload, using either Bayesian statistical modeling or machine learning approaches. Finally, the accuracy of the resulting workload detector will be evaluated when considering MATB more realistic paradigm.
|