Energy Distribution Image Processing of Stroke EEG Signal using Gray Level Co-occurrence Matrix

In this study, we propose the effect of Electroencephalography (EEG) stroke signal processing into energy distribution images using energy distribution for stroke conditions. The EEG signals are used as an alternative method to help the improvement of Brain Computer Interface (BCI) to detect stroke conditions. The energy distribution clarifies the relationship for each channel while converting the EEG signal into an energy distribution image. The Gray-Level Co-Occurrence Matrix (GLCM) with Genetic Algorithm (GA) and ANN-BP are used as a method for image feature values to get the optimal system feature after brain mapping using Power Spectrum Density (PSD). We evaluate the system performances via a series of computer simulations. We investigate the feature combination using GLCM by taking the best 11 features, i.e., contrast, correlation, variance, entropy, homogeneity, energy, sum variance, sum entropy, difference variance, difference entropy and inverse difference momentum with an accuracy equal to 61.25%. Thus, the GA uses to select the feature on GLCM in order to find the best combination for the BCI system in this study. We found the accuracy value of GA-GLCM equals 72.5% with features, i.e., contrast, correlation, homogeneity, energy, sum variance, and different variance, while the EEG signal is tested with accuracy equals 59%. The result shows that the BCI system can be optimized using the converted EEG signal into energy distribution images. The results are expected to contribute to the future of biomedical development.