CSE - Adaptive MI-BCI for Enhanced Neural Decoding

Project Overview:

Motor Imagery Brain-Computer Interfaces (MI-BCIs) using EEG are systems that allow users to control external devices or communicate by imagining specific movements, such as moving a hand or foot. This mental rehearsal generates distinctive patterns in the brain’s electrical activity, which are detected through Electroencephalography (EEG). MI-BCIs decode these patterns in real-time, enabling applications like neurorehabilitation, assistive technology, and interactive systems for individuals with motor impairments. As a non-invasive method, EEG-based MI-BCIs offer a practical approach to harnessing neural signals directly for intuitive human-computer interaction.

The proposed project aims to develop a novel MI-BCI using open-access EEG data. The goal is to achieve a significant enhancement in MI-BCI performance through the development of a new approach that leverages advanced signal processing and machine learning techniques. The research focuses on overcoming existing limitations in MI-BCI systems, such as low classification accuracy and user variability, by introducing a novel method that integrates domain adaptation and deep learning-based feature extraction. This project targets a publication in a high-impact journal, showcasing a novel approach that can improve MI-BCI usability in real-world applications.

Goals:

The primary objective of the research is to propose and validate a new method that improves the decoding of motor imagery tasks from EEG signals. The expected outcomes include:

• Development of a novel deep learning-based architecture for feature extraction from EEG signals.

• Introduction of a domain adaptation technique to improve generalization across different subjects and sessions.

• Demonstrating the effectiveness of the proposed approach using publicly available EEG datasets.

• Setting a new benchmark for MI-BCI performance that can be utilized in future studies.

Tasks and Techniques:

The project will be executed in the following phases:

1. Data Collection and Preprocessing:

• Utilize open-access MI-BCI datasets.

• Apply data preprocessing techniques to remove noise and artifacts from the raw EEG data.

2. Feature Extraction, Classification and Evaluation:

• Develop a novel deep learning-based approach for extracting discriminative features from the preprocessed EEG data to capture both spatial and temporal information.

• Compare the performance of traditional feature extraction techniques with the proposed deep learning approach.

• Evaluate the classification performance of the proposed system using metrics such as accuracy, precision, recall, and F1-score.

• Perform statistical analysis to validate the improvements over baseline methods.

3. Domain Adaptation:

• Introduce a domain adaptation technique to enhance the generalization of the trained models across different subjects and sessions, potentially using transfer learning or adversarial domain adaptation.

Student Participation:

Students involved in this project will gain hands-on experience in EEG signal processing, machine learning, and BCI system design. Their roles will include:

• Data Handling: Preprocessing EEG data and implementing noise removal techniques.

• Algorithm Development: Coding and testing new feature extraction and classification algorithms.

• Model Training and Evaluation: Training deep learning models and performing statistical evaluations of the results.

• Documentation and Dissemination: Contributing to writing the journal paper and presenting findings.

This project not only contributes to advancing the state-of-the-art in MI-BCI research but also provides an educational experience for students in computational neuroscience and machine learning.