Classification of Imagined Speech Signals Using Functional Connectivity Graphs and Machine Learning Models.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Brain-computer interfaces (BCIs) aim to decode neural signals for device control.
• Imagined speech classification using electroencephalography (EEG) is challenging due to signal complexity and variability.
• Current methods struggle with low signal-to-noise ratios and inter-subject differences.

Purpose of the Study:

• To develop a novel method for improved imagined speech classification in BCIs.
• To address the limitations of existing approaches in handling complex and noisy EEG data.
• To leverage functional brain connectivity for more accurate speech decoding.

Main Methods:

• Implementation of an imagined speech functional connectivity graph (ISFCG) to represent neural data.
• Extraction of graph-based features capturing relationships between brain regions during imagined speech.
• Application of a convolutional neural network (CNN) for feature learning and classification.

Main Results:

• The ISFCG method effectively captures complex brain interactions during imagined speech.
• The proposed CNN model, utilizing ISFCG features, achieved improved classification accuracy.
• Experimental validation on a benchmark dataset confirmed the method's efficacy.

Conclusions:

• The ISFCG approach offers a promising alternative for analyzing and classifying imagined speech signals.
• Focusing on functional connectivity enhances the robustness and accuracy of BCI systems.
• This method has the potential to advance the field of neural decoding for communication.