An Interface using Electrooculography with Closed Eyes

抄録

The purpose of our study was to determine the electrooculography (EOG) interface that can be used for amyotrophic lateral sclerosis (ALS) patients that can use without preparation. We proposed eye movement detection using Root Mean Square the active threshold (AT method) and k-nearest neighbor (k-NN) methods. The AT method is the threshold method that is dynamically calculated using the root mean square. This report describes the combined use of both methods. We conducted experiments on the waveform detection accuracy for 19 healthy subjects between the ages of 20 and 29 years. The hit rate for the proposed method was 94%, and the FA rate was 9%. Next, we calculated the information transfer rate (ITR), a popular evaluation index in for the brain-computer interface. (BCI). The minimum ITR using the proposed method was 19.02[bits/min]. The ITR of the P300 speller a BCI tool, was 16.4[bits/min]. The ITR of the proposed method was higher than that using the P300 speller. Therefore, it can be said that the proposed method has usefulness as an interface.