Improving EEG-based error detection using relative peak features.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) offer communication for severely disabled individuals.
• Accurate detection of error-related potentials (ErrPs) in EEG is crucial for BCI performance.
• Current ErrP detection methods have limitations in accuracy, especially for certain users.

Purpose of the Study:

• To propose a novel relative peak feature (RPF) selection method for improved ErrP detection in EEG.
• To enhance the accuracy and performance of BCIs by addressing challenges in ErrP recognition.
• To reduce the number of features required for accurate ErrP detection.

Main Methods:

• A new relative peak feature (RPF) selection method was developed and applied to EEG data.
• Data was collected from 29 participants (mean age 24.14 years).
• Classifiers including K-Nearest Neighbors (KNN), Support Vector Machines (SVM), and Linear Discriminant Analysis (LDA) were used.

Main Results:

• The RPF method achieved an average accuracy of 81.63% for detecting erroneous events and 78.87% for correct events across classifiers.
• Compared to temporal feature selection, RPF showed a 17.85% gain in error detection accuracy.
• RPF significantly reduced the number of features by 91.7% compared to state-of-the-art temporal features.

Conclusions:

• The proposed RPF method significantly improves the accuracy of ErrP detection in EEG signals.
• This advancement has the potential to enhance human-robot interaction by enabling more reliable error correction in BCIs.
• The RPF method offers a more efficient approach to feature selection for ErrP detection.