OBJECTIVE

Brain-computer interfaces based on steady-state visual evoked potentials (SSVEP-BCIs) achieve the highest performance, due to their multiclass nature, in paradigms in which different visual stimuli are shown. Studies of independent binary SSVEP-BCIs have been previously presented in which it was not necessary to gaze at the stimuli at the cost of performance. Despite that, the energy of the SSVEPs is largely affected by the temporal and spatial frequencies of the stimulus, there are no studies in the BCI literature about its combined impact on the final performance of SSVEP-BCIs. The objective of this study is to present an experiment that evaluates the best configuration of the visual stimulus for each subject, thus minimizing the decline in performance of independent binary SSVEP-BCIs.

METHODS

The participants attended and ignored a single structured stimulus configured with a combination of spatial and temporal frequencies at a time. They were instructed to gaze at a central point during the whole experiment. The best combination of spatial and temporal frequencies achieved for each subject, in terms of signal-to-noise ratio (SNR), was subsequently determined.

RESULTS

The SNR showed a significant dependency on the combination of frequencies, in such a way that only a reduced set of these combinations was applicable for obtaining an optimum SNR. The selection of an inappropriate stimulus configuration may cause a degradation of the information transmission rate (ITR) as it does the SNR.

CONCLUSIONS

The appropriate selection of the optimal spatial and temporal frequencies determines the performance of independent binary SSVEP-BCIs. This fact is critical to enhance its low performance; hence, they should be adjusted independently for each subject.

SIGNIFICANCE

Independent binary SSVEP-BCIs can be used in patients who are unable to control their gaze sufficiently. The correct selection of the spatial and temporal frequencies has a considerable benefit on their low ITR that must be taken into account. In order to find the most suitable frequencies, a test similar to the presented in this study should be performed beforehand for each SSVEP-BCI user. This regard is not documented in studies conducted in the BCI literature.