EEG Signals for Upper Limb Motor Imagery Classification

This dataset contains electroencephalographic (EEG) signals collected from 12 healthy individuals for a brain-computer interface (BCI) study. It supports a 3-class classification task discriminating rest against imaginary grasps and elbow movements of the same upper limb. The proposed time-domain feature extraction and SVM classification method achieved an average accuracy of 74.2% on this data.

Use Cases

• Train a support vector machine (SVM) classifier to discriminate between EEG signals of rest, imaginary grasp, and imaginary elbow movement states.
• Benchmark time-domain feature extraction methods against other approaches like common spatial patterns (CSP) or band power for motor imagery BCI accuracy.
• Analyze EEG signal patterns specific to imaginary movements within the same limb, which have close spatial representations on the motor cortex.
• Develop models for BCI-driven robotic devices, such as portable exoskeletons, using EEG data from healthy individuals to assist those with impaired upper extremity function.

Strengths

• Data is from a controlled study involving 12 healthy human subjects.
• The associated classification method achieved a benchmarked accuracy of 74.2%, outperforming CSP, FBCSP, and band power methods on this data.
• Focuses on the challenging task of classifying three distinct motor imagery states within the same limb.

Limitations

• The sample size is limited to 12 individuals, which may restrict the generalizability of models trained on this data.
• Data was collected prior to the described study, and specific details on recording parameters, preprocessing, or raw signal availability are not provided.
• The dataset's applicability is specific to upper limb motor imagery and may not generalize to other BCI paradigms or body parts.

Provenance

Source

Dryad digital repository, authored by Mojgan Tavakolan.

Collection Method

EEG signals were collected from 12 healthy individuals for a BCI study prior to the described research.

Freshness

Last updated in June 2020.