Related Experiment Video
Updated: Jun 7, 2025
09:42
Author Spotlight: Using Motor Imagery Brain-Computer Interface to Improve Motor and Cognitive Function in Stroke Patients
Published on: September 1, 2023
1.1K
Boosted Harris Hawks Shuffled Shepherd Optimization Augmented Deep Learning based motor imagery classification for
Fatmah Yousef Assiri1, Mahmoud Ragab2
1Software Engineering Department, College of Computer Science and Engineering, University of Jeddah, Jeddah, Saudi Arabia.
Plos One
|November 21, 2024
Summary
This study introduces a novel deep learning approach for motor imagery classification in brain-computer interfaces. The Boosted Harris Hawks Shuffled Shepherd Optimization Augmented Deep Learning (BHHSHO-DL) technique significantly improves accuracy for assistive technologies.
Area of Science:
- Neuroscience and Artificial Intelligence
- Brain-Computer Interfaces (BCI)
- Machine Learning for Neural Signal Processing
Background:
- Motor imagery (MI) classification is crucial for brain-computer interfaces (BCIs), enabling individuals with motor impairments to control external devices.
- Current BCIs often utilize electroencephalography (EEG) and machine learning (ML) for interpreting brain activity patterns during MI tasks.
- Advanced deep learning (DL) models are increasingly employed to enhance the accuracy and robustness of MI classification.
Purpose of the Study:
- To present a novel Boosted Harris Hawks Shuffled Shepherd Optimization Augmented Deep Learning (BHHSHO-DL) technique for motor imagery classification in BCIs.
- To leverage hyperparameter-tuned deep learning for improved MI identification and BCI performance.
- To enhance communication and mobility for individuals with motor disabilities through advanced BCI technology.
Main Methods:
- Data preprocessing using Wavelet Packet Decomposition (WPD).
- Feature extraction via enhanced DenseNet (Densely Connected Networks).
- Hyperparameter optimization using Boosted Harris Hawks Shuffled Shepherd Optimization (BHHSHO).
- Classification using Convolutional Autoencoder (CAE).
Main Results:
- The BHHSHO-DL methodology achieved superior classification accuracy on benchmark datasets.
- Achieved 98.15% accuracy on the BCIC-III dataset and 92.23% on the BCIC-IV dataset.
- Demonstrated significant performance improvements over existing techniques.
Conclusions:
- The BHHSHO-DL technique offers a highly accurate and effective method for motor imagery classification in BCIs.
- This advanced approach holds promise for improving the functionality and usability of BCIs for assistive purposes.
- The study highlights the potential of integrating metaheuristic optimization with deep learning for complex neural signal analysis.
