The Design of Stroke EEG Channel Selection System Using Spatial Selection Method

Stroke can be interpreted as a dysfunction of the nervous system that occurs suddenly and caused by blockage of blood vessels in the brain. Generally, the effort used to reduce stroke patients is the diagnostic method using Magnetic Resonance Imaging (MRI). However, the cost of examination using the MRI method is relatively expensive and not portable. One solution to overcome this problem is to use an Electroencephalograph (EEG) device to detect stroke signals in the brain that measure electrical activity detecting abnormalities in the brain. This action uses special sensors, namely electrodes attached to the head and connected to the computer. In previous research, EEG stroke signal processing was carried out using the Brain Symmetry Index and Hilbert Huang Transform (BSI-HHT) methods. However, this study did not specifically discuss channel selection in EEG stroke signals. Given these problems, in this study, the authors will process the EEG stroke signal using the modified Spatial Selection method using the Fast Fourier Transform (FFT) method through the active channel composition configuration so that it can be processed to obtain relevant results. Furthermore, the classification process is carried out using the k-Nearest Neighbor (k-NN) and Extreme Learning Machine (ELM) methods. Implementing the k-Nearest Neighbor (k-NN) classification shows that the spatial selection method can find the suitable channel composition with the same accuracy results as normal data in several areas. In contrast, the ELM classification can increase accuracy by 2% greater than normal data in the high mean area with a few channel compositions.