Recognition of cognitive load and resting state in motion using mobile optical tomography (DOT-fNIRS)

This project focuses on developing a real-time classification algorithm to distinguish between high cognitive-load and relaxation states during physical movement. By leveraging high-density mobile functional Near-Infrared Spectroscopy (fNIRS), the study monitors brain activity in participants (N = 60) while they are stationary or walking. The goal is to develop a model that identifies mental fatigue or recovery phases based on hemodynamic responses in specific cortical regions.

Hypothesis

The project is built on the physiological premise that distinct neural patterns emerge during mental effort versus rest, even when the subject is physically active:

1. Cognitive Load: Intense mental tasks (mental calculations) will trigger significant activation—evidenced by an increase in oxygenated hemoglobin (HbO)—in the dorsolateral prefrontal cortex (dlPFC). Simultaneously, deactivation will occur in the medial prefrontal cortex (mPFC), a hub of the Default Mode Network (DMN).
2. Relaxation: The recovery phase will exhibit the inverse pattern: activation of the mPFC (DMN) and a decrease in activity within the dlPFC.

Equipment Used

The study utilises mobile diffuse optical tomography near-infrared spectroscopy (DOT-fNIRS). Specifically, a high-density mobile fNIRS device is employed, enabling high-resolution monitoring of brain oxygenation while the user is in motion (e.g., walking on a treadmill) without the constraints of traditional stationary imaging.

Experimental Design and Procedures

The dataset includes 60 participants across a wide age range (18–76 years), ensuring demographic diversity. The classification model is trained on data from two primary conditions:

1. Single-Task (ST): Participants perform a complex mental task (multiplying two-digit numbers) or remain in a resting state while seated.  
2. Dual-Task (DT): Participants perform the same mental task and resting phases, but while walking on a treadmill at a self-adjusted pace.

Implementation Goal

The final output is a predictive model (e.g., neural network). This model processes real-time changes in HbO and deoxygenated hemoglobin (HbR) to determine when an individual is under intense mental strain or in a state of regeneration. This may have significant implications for workplace safety, sports science, and neuroergonomics, where monitoring mental state during physical exertion is critical.