Related Experiment Video
Updated: Dec 4, 2025
07:37
Assessment and Communication for People with Disorders of Consciousness
Published on: August 1, 2017
9.5K
BCIAUT-P300: A Multi-Session and Multi-Subject Benchmark Dataset on Autism for P300-Based Brain-Computer-Interfaces
Marco Simões1,2, Davide Borra3, Eduardo Santamaría-Vázquez4,5
1Coimbra Institute for Biomedical Imaging and Translational Research (CIBIT), Institute of Nuclear Sciences Applied to Health (ICNAS), University of Coimbra, Coimbra, Portugal.
Frontiers in Neuroscience
|October 26, 2020
Summary
A new dataset of Brain-Computer Interface (BCI) P300 signals from individuals with autism spectrum disorder provides a valuable resource for developing advanced BCI algorithms. This comprehensive dataset enables research into deep learning methods for improved BCI performance.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Computer Science
Background:
- Limited availability of multi-session P300 datasets hinders Brain-Computer Interface (BCI) research.
- Existing public datasets often lack sufficient participants and sessions, impeding the development of advanced data processing and analysis methods.
- The need for large datasets is critical for exploring deep learning applications in BCI systems.
Purpose of the Study:
- To introduce and characterize the BCIAUT-P300 dataset, a novel resource for P300-based BCI research.
- To provide a benchmark dataset for evaluating BCI algorithms, particularly those utilizing multiple sessions.
- To facilitate the advancement of BCI technologies, especially for individuals with autism spectrum disorder.
Main Methods:
- Collected electroencephalography (EEG) data from 15 individuals with autism spectrum disorder over 7 sessions each, totaling 105 sessions.
- Utilized a P300-based BCI joint-attention training paradigm.
- Organized the 2019 IFMBE Scientific Challenge using the dataset to evaluate various BCI algorithms.
Main Results:
- The BCIAUT-P300 dataset comprises 105 sessions from 15 participants with autism spectrum disorder.
- The 2019 IFMBE Scientific Challenge demonstrated the dataset's utility, with the winning team achieving 92.3% accuracy using a convolutional neural network (EEGNet).
- The dataset is now publicly available for further research and development.
Conclusions:
- The BCIAUT-P300 dataset addresses the critical need for multi-session P300 data in BCI research.
- This resource is expected to accelerate the development of more effective BCI algorithms, including those based on deep learning.
- The dataset serves as a valuable benchmark for future P300-based BCI studies.
