Bayesian delay time estimation of brain signal using N100 response for auditory BCI

Brain computer interface (BCI) enables disabled people to communicate by brain signal. P300 which appears 300ms after the onset of a low frequent stimulus is extensively used to actualize BCI. Precise detection of P300 component is therefore important. Most of existing BCI assumes that P300 is observed after 300ms, however this latency has variation due to the condition of a subject and the level of attention for the stimulus. This latency variation distorts averaged P300 and hence incurs the deterioration of the classification accuracy. A delay time estimation method for P300 signal using Bayesian estimation has been reported in the previous study to address this problem. However, the method has a problem that the algorithm fails to estimate the delay time when the signal does not contain P300. A Bayesian delay time estimation method using N100 component is therefore proposed. This proposed method exhibited 3.2% higher classification accuracy than the conventional delay time estimation method in auditory BCI.