Identifying EEG Parameters to Monitor Stroke Rehabilitation using Individual Analysis

Monitoring post-stroke patients is significantly important to determine the progress of activity in the brain during the rehabilitation period. The monitoring method using electroencephalograph (EEG) can provide more objective and measurable results. One of the challenges in this field is to determine the right EEG parameters. Most of the previous studies focus more on test selected parameters with one condition of stroke, hemiparesis, or hemiplegia only, while in this study we focused not only on the test selected parameters, but also choosing the most stable parameters, band frequency, movement, and identify the relation between stroke conditions with parameter values. In this experiment, we compared healthy hand movements (HHM) against affected hand movements (AHM) in individual post-stroke patients using several EEG features at varying frequencies and motion. More details, we measured the brain activity of 10 respondents through 2 EEG electrodes (C3 and C4) when performing 3 active motor task: grasping, elbow flexion-extension, and shoulder flexion-extension. We use the Artifact Subspace Reconstruction (ASR) method to cleaning noise artifact from EEG data. Furthermore, we have separated EEG data into three band frequencies: alpha, beta low, and beta high. The features that apply are standard deviation (STD), mean absolute value (MAV), power spectral density (PSD), and power percentage (PP). Then, we use the individual analysis to calculate the difference value between HHM against AHM on each feature, band frequency, and movements. The result showed that among four parameters (STD, MAV, PSD, and PP), PSD reached the highest difference value in the alpha band when comparing between HHM and AHM. When comparing three motions, grasping motion showed the most significant difference between HHM and AHM. We also found that the Hemiplegia patients showed lower value in all parameters in AHM when comparing with Hemiparesis patients. The conclusion is certain task motion, and certain parameter needs to be applied when monitoring the progress of stroke patient rehabilitation.