simBCI - A framework for studying BCI methods by simulated EEG

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computer Science

Background:

• Brain-Computer Interface (BCI) methods often rely on Electroencephalogram (EEG) data from human experiments.
• Acquiring real EEG data is costly and lacks a known generative ground truth due to unobservable brain mechanisms.
• Existing methods face challenges in understanding and developing BCI signal processing techniques.

Purpose of the Study:

• Introduce simBCI, a novel framework for generating artificial BCI data.
• Provide a controlled environment for testing and studying BCI signal processing methods.
• Facilitate the development and debugging of BCI classification pipelines.

Main Methods:

• Developed a framework for simulating BCI data, enabling controlled experimentation.
• Implemented open-source models and methods for data generation and signal processing.
• Allowed for iteration over models and parameters for data generation and processing.

Main Results:

• Created a convenient framework, simBCI, for generating artificial BCI data.
• Enabled testing of classification pipelines using simulated data under controlled conditions.
• Provided open-source implementations of various BCI models and methods.

Conclusions:

• simBCI offers a valuable tool for hypothesis discovery, algorithm study, education, and pipeline debugging in BCI research.
• The framework is modular, extensible, and freely available as open-source software.
• simBCI complements, rather than replaces, human experiments in BCI research.