Framework
BCI Essentials uses Lab Streaming Layer (LSL) to interchange time synchronized EEG data, presentation markers, and classification responses. A stream with type: BCI_Essentials_Markers sends markers from the front end. Another with type: BCI_Essentials_Predictions sends predictions from the back end. Messages are formatted as comma-delimited strings.
| Stream Type | Content |
|---|---|
BCI_Essentials_Markers | Data Markers |
BCI_Essentials_Predictions | Classifier Inferences |
Timeline Definitions
- Epoch
- Data processing window surrounding a period of focus on a target class or object. For discrete stimulus paradigms like P300, an epoch consists of a baseline period followed by an expectation of stimulus response. Paradigms targetting a specific state such as MI and SSVEP define these classed periods arbitrarily in conjunction with user prompting.
- Trial
- Stimulus sequence or set of epochs with user focus on a particular class or object. Used for both training and classification.
- Training
- Sequence of targetted trials targetting different objects or classes providing the data and event markers necessary to build a classifier for live inferences.
- An Epoch must contain enough data to meaningfully classify; at least 1-2 seconds
- All training and classification Epochs must have the same length
- Epochs may overlap
- A Trial is demarcated by event markers
- P300 Trials contain multiple “flashing” stimulus events
- SSVEP or Motor Imagery Trials consist of one or more Epochs
- A Training Sequence contains one or more Trials
