An Intelligent Motor Imagery Detection System Using Electroencephalography with Adaptive Wavelets.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Brain-computer interfaces (BCIs) enable communication for individuals with disabilities.
• Accurate electroencephalography (EEG) signal processing is crucial for effective BCIs.
• Optimizing parameters for signal decomposition methods like tunable Q wavelet transform (TQWT) is challenging.

Purpose of the Study:

• To propose a robust TQWT method for automatic optimization of tuning parameters.
• To accurately decompose non-stationary EEG signals for motor imagery (MI) tasks.
• To enhance the performance of BCIs for specially-abled individuals.

Main Methods:

• Developed a robust TQWT approach for automated parameter selection.
• Employed three evolutionary optimization algorithms to find optimal TQWT parameters.
• Utilized Laplacian score for dominant EEG channel selection.
• Classified extracted features using least square support vector machine (LS-SVM) with various kernels.

Main Results:

• Achieved the highest classification accuracy of 99.78% using the radial basis function kernel.
• Demonstrated the superiority of the proposed method over existing state-of-the-art techniques on the same dataset.
• Successfully decomposed non-stationary EEG signals with high precision.

Conclusions:

• The proposed robust TQWT with automated parameter tuning significantly improves EEG signal decomposition for MI tasks.
• The method offers a superior and accurate approach for BCI applications.
• This advancement holds promise for enhancing BCIs for individuals with motor impairments.