Multi-Feature Fusion Method Based on EEG Signal and its Application in Stroke Classification.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalogram (EEG) analysis is a cost-effective, noninvasive tool for diagnosing stroke.
• Accurate classification of stroke types, specifically cerebral infarction (ischemic) and cerebral hemorrhage (hemorrhagic), is crucial for effective treatment.
• Existing entropy-based methods for EEG signal analysis show varying effectiveness in characterizing stroke patient data.

Purpose of the Study:

• To propose a novel EEG stroke signal classification method for differentiating between ischemic and hemorrhagic stroke.
• To introduce a multi-feature fusion approach combining wavelet packet energy, fuzzy entropy, and hierarchical theory.
• To construct an optimal ensemble classification model for stroke EEG signals.

Main Methods:

• A multi-feature fusion method was developed by integrating wavelet packet energy and hierarchical fuzzy entropy.
• Hierarchical fuzzy entropy was extracted by combining hierarchical theory with fuzzy entropy, demonstrating superior performance over other entropy measures (permutation, sample, approximate) for stroke EEG signals.
• Ensemble classifiers, including Support Vector Machine (SVM), decision tree, and random forest, were employed for classification.

Main Results:

• The proposed multi-feature fusion method, combining wavelet packet energy and hierarchical fuzzy entropy, significantly enhanced classification accuracy compared to using fuzzy entropy alone.
• The ensemble random forest model achieved the highest classification accuracy among the tested models when utilizing the fused features.
• The developed method proved efficient in classifying ischemic and hemorrhagic stroke based on EEG signals.

Conclusions:

• The novel multi-feature fusion approach integrating wavelet packet energy and hierarchical fuzzy entropy is effective for stroke EEG signal classification.
• The ensemble random forest classifier provides optimal performance for distinguishing between ischemic and hemorrhagic stroke using the proposed feature fusion.
• This method offers a promising advancement in the noninvasive diagnosis and classification of stroke types via EEG analysis.