Adaptive-projection intrinsically transformed multivariate empirical mode decomposition in cooperative brain-computer...

Area of Science:

• Signal Processing
• Biomedical Engineering
• Data Analysis

Background:

• Multivariate empirical mode decomposition (MEMD) is a signal processing technique.
• Real-world multichannel data often exhibit power imbalances and inter-channel correlations.
• Existing MEMD methods may struggle with these data characteristics.

Purpose of the Study:

• To introduce an extension of MEMD called adaptive-projection intrinsically transformed MEMD (APIT-MEMD).
• To evaluate APIT-MEMD's performance against MEMD, particularly with limited projection vectors.
• To apply APIT-MEMD within an intrinsic multiscale analysis framework for brain-computer interfaces (BCI).

Main Methods:

• Developed adaptive-projection intrinsically transformed MEMD (APIT-MEMD).
• Compared APIT-MEMD and MEMD performance using varying numbers of projection vectors.
• Integrated noise-assisted APIT-MEMD into an intrinsic multiscale analysis framework.
• Applied the framework to cooperative BCI paradigms using steady-state visual evoked potentials and P300 responses.

Main Results:

• APIT-MEMD shows comparable or superior performance to MEMD, especially with fewer projection vectors.
• The noise-assisted APIT-MEMD approach demonstrates advantages in noise-dominated BCI.
• The intrinsic multiscale analysis framework improved system performance in a joint cognitive BCI task.

Conclusions:

• APIT-MEMD is an effective extension for handling power imbalances and correlations in multichannel data.
• The proposed intrinsic multiscale analysis framework, utilizing APIT-MEMD, enhances BCI performance.
• This approach offers significant improvements for cooperative and cognitive BCI applications, particularly in noisy environments.