Investigating singing imagery as an additional or alternative control task for EEG-based Brain-Computer Interfaces.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Brain-computer interfaces (BCIs) enable communication and control without physical movement, typically using motor imagery (MI) tasks.
• Current BCIs often classify limited MI tasks (e.g., hand, foot, tongue), restricting command options and system utility.
• Expanding the repertoire of mental tasks is crucial for enhancing BCI functionality and user experience.

Purpose of the Study:

• To investigate the efficacy of singing imagery (SI) as a novel mental task for BCIs.
• To evaluate SI's performance as both an alternative and an additional task to conventional MI paradigms.
• To assess the impact of including SI on the number of achievable commands in multi-class BCI scenarios.

Main Methods:

• EEG data collected from 14 participants performing MI (hand, foot, tongue), SI, and rest conditions.
• Filter bank common spatial patterns (FBCSP) used for feature extraction.
• Random forest algorithm employed for classification across 2-, 4-, 5-, and 6-class settings.

Main Results:

• Classification accuracies incorporating SI were comparable to subsets of MI tasks in 2-, 4-, and 5-class scenarios.
• In a 6-class scenario, average accuracy reached approximately 60%, with some participants exceeding 70%.
• Subjective feedback on task difficulty and preferences was also gathered.

Conclusions:

• Singing imagery (SI) is a viable additional or alternative task for motor imagery-based BCIs (MI-BCIs).
• SI can increase the number of available commands, enhancing the functional capacity of BCI systems.
• This research advances the development of more flexible and user-centered BCI applications by broadening task options.