The Application of Superlet Transform in EEG-Based Motor Imagery Classification of Unilateral Knee Movement

The Brain-Computer Interface (BCI) establishes a direct communication pathway between the brain and external devices, enabling people to control external devices through EEG-based motor imagery. There have been numerous studies and applications of motor imagery of the upper limb in recent years. However, studies of motor imagery of the lower limb for BCI applications remain insufficient. This study aims to decode motor imagery tasks of unilateral knee extension and flexion, which could be used as input for rehabilitation robots or exoskeletons. Superlet is a spectral estimator with super-resolution time-frequency that has been recently proposed in the field of signal processing. In this study, the superlet transform was performed for time-frequency analysis of the EEG signals of unilateral knee motor imagery recorded from five healthy subjects. The extracted superlet coefficients from the theta, alpha, and beta bands were used as features. Subsequently, the SVM was used to build binary classification models for each subject. The results of the study show that the average offline classification accuracy achieved using the superlet and SVM is 78.32%, which is better than that achieved using the wavelet and SVM.