Decoding olfactory EEG signals for different odor stimuli identification using wavelet-spatial domain feature.

Area of Science:

• Neuroscience
• Signal Processing
• Biomedical Engineering

Background:

• Olfactory electroencephalography (EEG) signal decoding is crucial for applications like disease diagnosis and brain-computer interaction (BCI).
• Challenges include extracting discriminative features from low spatial resolution and poor signal-to-noise ratio olfactory EEG signals.
• Improving decoding accuracy is vital for advancing these applications.

Purpose of the Study:

• To develop a novel feature extraction method for decoding olfactory EEG signals.
• To enhance the accuracy of olfactory EEG signal classification.
• To address the challenges of low spatial resolution and poor signal-to-noise ratio in olfactory EEG.

Main Methods:

• A new feature, Wavelet-Spatial Domain Feature (WSDF), is proposed by combining Discrete Wavelet Transform (DWT) and One-Versus-Rest Common Spatial Pattern (OVR-CSP).
• DWT is used for multilevel decomposition of EEG signals.
• OVR-CSP is applied to DWT coefficients for spatial filtering, followed by variance extraction to generate the WSDF.

Main Results:

• The WSDF method achieved high classification accuracies: 100% (eyes-open) and 94.47% (eyes-closed) on the Odor Pleasantness Perception Dataset (OPPD).
• On a self-collected dataset, the WSDF method achieved a highest average accuracy of 99.50%.
• WSDF demonstrated superior classification performance compared to existing EEG features and methods on the same datasets.

Conclusions:

• The proposed Wavelet-Spatial Domain Feature (WSDF) is a highly effective method for decoding olfactory EEG signals.
• WSDF shows significant potential for improving accuracy in BCI and diagnostic applications.
• This novel feature extraction technique offers a promising solution for challenges in olfactory EEG signal analysis.