Spatial selection configuration on EEG stroke signal

Abstract

Stroke is one of the cerebrovascular health disorders caused by a blockage of blood flow to the brain. Data from South East Asian Medical Information Center (SEAMIC) explain that the most significant stroke mortality occurred in Indonesia, Philippines, Singapore, Brunei, Malaysia, and Thailand. There are several methods for diagnosing stroke, one of which is an electroencephalograph (EEG). EEG is one of the more widely used BCI methods due to its lower price, portability, ease of use, and high temporal resolution than other methods. Most EEG require signals from various places on the scalp to achieve good performance. However, using a large number of channels can degrade signal performance in the EEG. Spatial selection can be used for channel selection in the EEG stroke signal which can then be useful for evaluation of stroke therapy. Therefore, this study will select the optimized channel using the spatial selection method to see which channels are relevant to the EEG stroke signal. Also, using the Power Spectral Density extraction feature and Extreme Learning Machine classification. The L2-norm energy calculation method gets better results than other methods. This method can also select the active channel relevant to the stroke EEG signal. The results show that the spatial selection method can increase accuracy by 15 percent and optimize the system with 37.5 percent channel reductions.