EEG-based 90-Degree Turn Intention Detection for Brain-Computer Interface

TL;DR

The paper addresses pre-event turn-intention detection for lower-limb movement using EEG signals. It proposes a complete pipeline—from preprocessing and source localization to feature extraction with Hjorth-based and statistical features, feature selection via Random Forest, and classification with XGBoost, Gradient Boosting, and SVM—evaluated on data from nine participants collected with 31-channel EEG and IMU kinematics. The SVM with a window of $[-1.5,0]$ seconds achieves the best mean metrics, $ACC=81.23\%$, $PR=85.35\%$, and $REC=83.92\%$, and the study demonstrates feasibility of detecting turn intentions up to approximately $250$ ms before onset using pre-movement EEG. Key contributions include demonstrating pre-event decoding for left/right turns and straight walking, applying sLORETA for brain-source insights, and showing that a compact feature set (50 features) suffices after RF-based selection. The work has practical implications for proactive BCI control of assistive devices and gait rehabilitation, highlighting the potential for real-time, pre-movement intention decoding in mobile or non-laboratory settings.

Abstract

Electroencephalography (EEG)--based turn intention prediction for lower limb movement is important to build an efficient brain-computer interface (BCI) system. This study investigates the feasibility of intention detection of left-turn, right-turn, and straight walk by utilizing EEG signals obtained before the event occurrence. Synchronous data was collected using 31-channel EEG and IMU-based motion capture systems for nine healthy participants while performing left-turn, right-turn, and straight walk movements. EEG data was preprocessed with steps including Artifact Subspace Reconstruction (ASR), re-referencing, and Independent Component Analysis (ICA) to remove data noise. Feature extraction from the preprocessed EEG data involved computing various statistical measures (mean, median, standard deviation, skew, and kurtosis), and Hjorth parameters (activity, mobility, and complexity). Further, the feature selection was performed using the Random forest algorithm for the dimensionality reduction. The feature set obtained was utilized for 3-class classification using XG boost, gradient boosting, and support vector machine (SVM) with RBF kernel classifiers in a five-fold cross-validation scheme. Using the proposed intention detection methodology, the SVM classifier using an EEG window of 1.5 s and 0 s time-lag has the best decoding performance with mean accuracy, precision, and recall of 81.23%, 85.35%, and 83.92%, respectively, across the nine participants. The decoding analysis shows the feasibility of turn intention prediction for lower limb movement using the EEG signal before the event onset.

Figures (3)

• Figure 1: Experimental recording paradigm.
• Figure 2: Block diagram depicting the proposed framework for turn-intention.
• Figure 3: Brain source localization using sLORETA at different time stamps for (a) Left turn, (b) Right turn, and (c) Straight walk.