App details
RCP Speller
This app implements the classical P300-based speller proposed by Farwell and Donchin
This app implements the first P300-based BCI proposed by Farwell and Donchin. The system uses the row-column paradigm (RCP) to determine the intentions of the user. The RCP displays a matrix (i.e., speller) with several commands. The rows and columns of the matrix are randomly highlighted a predefined number of times (i.e., sequences). The user has to stare at the desired command and count the flashes (i.e., target stimuli), while ignoring the other stimuli (i.e., non-target stimuli). Target stimuli elicits visual ERPs in the EEG just after each flash. The signal processing stage detects the visual ERPs and determines the command. This BCI is also named as P300-based speller because of the P300 potential, which is the most prominent peak of visual ERPs elicited by target stimuli.
Details
The P300 Speller: the most classical BCI paradigm
This app implements the classical P300 speller based on the the row-column paradigm (RCP) proposed by Farwell and Donchin.
The RCP displays a matrix of commands whose rows and columns are highlighted in a random order. For each selection (i.e., trial), the user has to stare at the target command while ignoring the other stimuli. When the trial ends, the application finds out the command by detecting the P300 elicited in the EEG of the user just after each target stimulus.
The original purpose of the RCP speller was to improve the independence and quality of life of severely disabled people by providing a new channel of communication between the brain and the environment . Although this paradigm has been surpassed in recent years in terms of precision and selection speed by more advanced options, such as SSVEPs and c-VEPs, it is still widely used and it can be applied to investigate attention, visual information processing and cognitive responses within the brain.
The RCP speller app for MEDUSA© Platform provides advanced options. First, it allows to configure all the important parameters: stimulus duration, inter-stimulus interval, text or icon commands, flashing colors or command functions. Moreover, the available models for P300 detection include widely used options, such as rLDA, or advanced deep convolutional neural networks, such as EEGNet or EEG-Inception to improve performance [2]. It is also worth mentioning that it allows to include nested matrices to design complex real-life applications. The current implementation has been used in several studies to date, delivering state-of-the-art results [1, 2, 3, 4].
Visit the app channel in the discord server to ask questions, report issues and make improvement suggestions!
The following GIF shows the application:
References:
[1] E. Santamaría-Vázquez, V. Martínez-Cagigal, J. Gomez-Pilar, R. Hornero, Asynchronous Control of ERP-Based BCI Spellers Using Steady-State Visual Evoked Potentials Elicited by Peripheral Stimuli, IEEE transactions on neural systems and rehabilitation engineering 735 27 (9) (2019) 1883–1892.
[2] E. Santamaría-Vázquez, V. Martínez-Cagigal, F. Vaquerizo-Villar, R. Hornero, EEG-Inception: A Novel Deep Convolutional Neural Network for Assistive ERP-based Brain-Computer Interfaces, IEEE Transactions on Neural Systems and Rehabilitation Engineering 28 (12) (2020) 2773–2782.
[3] V. Martínez-Cagigal, E. Santamaría-Vázquez, R. Hornero, Asynchronous Control of P300-Based Brain–Computer Interfaces Using Sample Entropy, Entropy 21 (3) (2019) 230.
[4] E. Santamaría-Vázquez, V. Martínez-Cagigal, S. Pérez-Velasco, 770 D. Marcos-Martínez, R. Hornero, Robust Asynchronous Control of ERP-Based Brain-Computer Interfaces using Deep Learning, Computer Methods and Programs in Biomedicine 215 (2022) 106623.
RCP Speller Tutorial
Welcome to the Row-Column Paradigm (RCP) Speller tutorial. Below are the RCP app configuration parameters, allowing you to customize the application settings according to your study requirements.
Initial Setup and Requirements
Before start, make sure that you already have the application installed in Medusa Platform (refer to the Apps panel for instructions on adding applications in Medusa)
RCP Speller Settings
The configuration panel includes fields for capturing user information, session details, and the speller configuration settings desired.
 |
Run Settings
|
 |
Timings
|
|
|
Color settings In these parameters you can customize the color of the commands and text into the matrix.
Sizes This parameter allow you to customize the size of:
Matrices With these parameters you can customize dimension of the matrix. |
Train models
After selecting Train Models within Run Settings, the following parameters will appear, allowing you to configure the model trainer to suit your data.
|
|
Data |
|
|
Dataset |
|
|
ERP Model Here, you can select the model type rLDA, EEGNet, or EEG-Inception and customize each model’s settings.
After clicking Train, the Model Trainer Log box will display the model's performance metrics of training set. To get this performance, an average score is computed across rows and columns for each epoch based on the scores assigned by the model (e.g. rLDA), indicating the probability of containing an ERP. Afterward, the command corresponding to the row and column with the highest score is chosen.
|
This tutorial includes all you need to do in order to set up the RCP Speller application. We also hope this is useful and efficient for you. Should you have any problems, or if you do have questions or even feedback, let us know through our Community.