Learning to control an SSVEP-based BCI speller in naïve subjects.

High-speed steady-state visual evoked potential (SSVEP)-based brain-computer interface (BCI) has been demonstrated in several recent studies. This study aimed to investigate some issues regarding feasibility of learning to control an SSVEP-based BCI speller in naïve subjects. An experiment with new BCI users was designed to answer the following questions: (1) How many people can use the SSVEP-BCI speller? (2) How much time is required to train the user? (3) Does continuous system use lead to user fatigue and deteriorated BCI performance? The experiment consisted of three tasks including a 40-class BCI spelling task, a psychomotor vigilance test (PVT) task, and a test of sleepiness scale. Subjects' reaction time (RT) in the PVT task and the fatigue rank in the sleepiness scale test were used as objective and subjective parameters to evaluate subjects' alertness level. Among 11 naïve subjects, 10 of them fulfilled the 9-block experiment. Four of them showed clear learning effects (i.e., an increasing trend of classification accuracy and information transfer rate (ITR)) over time. The remaining subjects showed stable BCI performance during the whole experiment. The results of RT and fatigue rank showed a gradually increasing trend, which is not significant across blocks. In summary, the results of this study suggest that controlling an SSVEP-based BCI speller is in general feasible to learn by naïve subjects after a short training procedure, showing no clear performance deterioration related to fatigue.

Developing a one-channel BCI system using a dry claw-like electrode.

An eight-class SSVEP-based BCI system was designed and demonstrated in this study. To minimize the complexity of the traditional equipment and operation, only one work electrode was used. The work electrode was fabricated in our laboratory and designed as a claw-like structure with a diameter of 15 mm, featuring 8 small fingers of 4mm length and 2 mm diameter, and the weight was only 0.1g. The structure and elasticity can help the fingers pass through the hair and contact the scalp when placed on head. The electrode was capable to collect evoked brain activities such as steady-state visual evoked potentials (SSVEPs). This study showed that although the amplitude and SNR of SSVEPs obtained from a dry claw electrode was relatively lower than that from a wet electrode, the difference was not significant. This study further implemented an eight-class SSVEP-based BCI system using a dry claw-like electrode. Three subjects participated in the experiment. Using infinite impulse response (IIR) filtering and a simplified threshold method based on fast Fourier transform (FFT), the average accuracy of the three participants was 89.3% using 4 sec-long SSVEPs, leading to an average information transfer rate (ITR) of 26.5 bits/min. The results suggested the ability of using a dry claw-like electrode to perform practical BCI applications.

A sapphire based monolithic integrated optrode.

A novel kind of optrode fabricated on a sapphire substrate is proposed for optogenetic applications in neuroscience. Eight thin-film neural electrodes and a GaN-LED are monolithically integrated on the surface of a sapphire shank. The LED is used for optogenetic stimulation and the multiple electrodes are used for simultaneous recording of neural activities. The output power density of the LED is 1-19 mW/mm2 at 468 nm, driving with a current from 0.7-10 mA. The mean electrochemical impedance of the eight recoding sites on the optrode at 1 kHz is 385 kΩ. The highest temperature-raise at tissue around the LED is almost 1 °C when the output power density is 3 mw/mm2. The monolithic integrated structure will make it a powerful tool for optogenetics.