Table 1.

Summary of the proposed method.

Input: Contaminated EEG data, vertical EOG, horizontal EOG
Output: Artifact-free EEG data
1. Decompose contaminated EEG data using ICA to get ICs
2. Calculate composite multi scale entropy and kurtosis to identify ocular artifact related ICs
3. Apply median absolute deviation to remove high magnitude ocular artifacts
4. Filter ICs with linear regression model and extended recursive least squares
5. Artifact-free EEG data by back projecting all ICs using inverse ICA