Robust Virtual Keyboard for Brain-Computer Interface (ROBIK): An Halfway Update on the Project

International audienceThe principle of a Brain-Computer Interface or BCI is to control a device through the extraction and interpretation of signal features from electroencephalograms (EEG) collected either from the surface of the scalp or through invasive measurements. This late idea of communication technique (Vidal 1973), offers the advantage of bypassing the need for muscle activity in the control chain and is therefore presented as a promising alternative to restore communication and control in severely disabled patients (Wolpow, et al. 2002). However, the lack of robustness and ergonomics of both available software and EEG measurement techniques have delayed the transfer of this technology to patients bedsides. The French Research Agency has funded a 3-year project gathering national leaders in microelectronics (CEA-Leti), EEG signal processing (Gipsa-Lab) and clinical management of severely disabled people (Raymond Poincar hospital). The aim of the project is the development and the clinical validation of a Brain-Computer Interface prototype for communication. As an initial step, a survey was carried out to assess patients' and users (family and caretakers) needs, which were translated into specifications, on the basis of which software and hardware were developed. The survey (n=45) highlighted the need for easy-to-setup systems (installation time=15min), which stresses the importance of mechanical comfort and customization of application. The development of signal processing techniques has led to improvements of the P3Speller paradigm. A first prototype of a 32-channel EEG recording system is under development. To ease the EEG measurements and reduce installation time, the system has a reduced size. It includes the analog amplification and digital conversion of 32 channels sampled at 1 kHz, as well as the wireless data transmission to a computer. First in vivo validations were performed on small animals. This system will be optimized and connected to a headset specifically designed to provide a comfortable and handy interface with dry electrodes. The present project will still run for one and a half years ,ending with its clinical validation in a population of severely disabled patients, which will compare performances of the system with existing assistive technologies. At this stage, the proposed system yields very promising results, and outperforms the current state-of-the-art. If such a system is shown to perform better than current users assistive technology, it could reach the commercial availability for severely disabled patients within the next 5 years