We have developed an operational prototype of a modular sensing platform for wearable electroencephalography (EEG) recordings and validated it in several brain–computer interface (BCI) paradigms, including auditory attention decoding. The platform is designed as a wireless EEG sensor network, consisting of multiple miniaturized wireless EEG sensor nodes that synchronously collect data from different scalp locations. With no wires between the sensors, the system allows flexible placement and is discreet in appearance. This stands in contrast to commercial headset-based EEG systems that have fixed electrode configurations and are often bulky. The absence of interconnecting wires also reduces sensitivity to motion artefacts and electromagnetic interference. While commercial systems may offer advantages in certain hardware metrics such as battery capacity or transmission range, our platform is optimized for miniaturization, scalability and modularity.
During the development process we first explored flexible polymer micro-needle electrodes, but these did not provide sufficiently low and stable contact impedance in our validation tests. We therefore redesigned the electrodes using more rigid silicon-based micro-needles with nanostructuring to improve contact impedance.
All components have been integrated into a single sensor node, with a form factor designed for easy attachment and removal. A flexible printed circuit board (flex-PCB) was developed that can be clicked onto adhesive electrodes. The EEG sensor PCB can be plugged into a connector embedded in the flex-PCB. Our system is also compatible with commercial pre-gelled disposable electrodes, enabling quick and efficient short-term setups. The sensor nodes also feature an embedded microphone, which is useful for analyzing neural responses to auditory signals (see below).
The system is described in the following publication:
R. Ding, C. Hovine, P. Callemeyn, M. Kraft and A. Bertrand, "A wireless, scalable and modular EEG sensor network platform for unobtrusive brain recordings", IEEE Sensors Journal, vol. 25, no. 2, 2025, pp. 22580–22590. doi: 10.1109/JSEN.2025.3562791.
The platform was tested in an auditory attention decoding experiment in which participants focused on a target speaker within a two-speaker mixture. Using EEG data collected jointly from multiple miniature sensor nodes, we successfully decoded which speaker the participant was attending to. The results showed that combining multiple sensors significantly improves decoding accuracy compared to a single EEG node. These findings will be presented in a forthcoming publication.
To demonstrate the versatility of the platform, we also tested it in three other brain–computer interface paradigms: steady-state visually evoked potentials, auditory steady state responses, and neural tracking of speech for hearing assessment.
