Simplify Research, Amplify Knowledge: Advancing BCI, Neurofeedback, and Cognitive Applications with MEDUSA

Introduction

Brain-computer interfaces (BCIs) have opened new possibilities in neuroscience. However, BCI research requires sophisticated software tools for signal acquisition, real-time processing, and experiment design. Many investigation groups face challenges due to the complexity of existing solutions, which can slow progress and limit accessibility.

MEDUSA, an open-source software platform, addresses the difficulties of designing and implementing BCI experiments, making BCI research more accessible and versatile. 

This post will explore applications including neurofeedback, BCI-driven communication and device control, and cognitive neuroscience and how researchers can leverage the platform to streamline their studies, reduce technical workload, and focus more on data analysis and interpretation.

What are Brain-Computer Interfaces?

Brain-computer interfaces (BCIs) are an innovative technology that enables direct communication between the human brain and external devices, bypassing traditional neuromuscular pathways. In a nutshell, this technology interprets neural activity to decode the user’s intentions and uncover underlying brain processes. BCIs can be categorized into invasive, minimally invasive, and non-invasive systems based on how neural signals are acquired. Non-invasive methods, such as electroencephalography (EEG), are the most widely used due to their accessibility and safety, allowing the creation of practical and scalable solutions.

Initially developed to assist individuals with severe motor impairments, BCIs empower users to control computers, robotic systems, and other devices in real time. Over time, the applications of this technology have expanded into a wide range of fields, including neurorehabilitation, cognitive training, and entertainment. 

Beyond these practical applications, BCIs have become important tools for advancing fundamental brain research, offering insights into neural processes, cognition, and consciousness. By enabling precise measurements and real-time manipulation of brain activity, they allow researchers to study the mechanisms of thought, perception, memory, and self-awareness. As the field continues to evolve, BCIs hold the potential to revolutionize healthcare, enhance human-machine interaction, and open new frontiers in neuroscience research.

The Role of Software Platforms in BCI Research

The rapid development of BCI technology has enabled groundbreaking insights and innovations in neuroscience. This interdisciplinary domain drives the design of increasingly sophisticated experiments that demand advanced software tools capable of executing and managing complex research paradigms with precision and efficiency. Such software must handle critical processes, including signal acquisition and synchronization, real-time data processing, task presentation, and user feedback.

Designing and developing these tools is a complex and time-consuming process that demands significant technical expertise. Unfortunately, many research groups lack the necessary skills or resources, often depending on limited off-the-shelf software or external developers. The consequences: project delays, higher costs, and technical challenges that ultimately hinder innovation and lead to repetitive, less impactful studies.

In this context, software tools specifically designed for neurotechnology research are critical for advancing a wide range of fields, as they can accelerate experimentation, lower costs, and empower clinical researchers without technical expertise. However, while many signal processing toolboxes exist, options for customizable experimental design and implementation are limited. The platform MEDUSA was created out of this necessity to provide a versatile, open-source platform that bridges these gaps. Below are the key features, advantages, and use cases of this software.

Overview of MEDUSA

MEDUSA© is an open-source software ecosystem developed to simplify the creation and execution of BCI experiments. It combines robust signal acquisition and processing tools with an intuitive user interface, enhancing accessibility to BCI research. With a design based on modularity, flexibility, and scalability, fits a broad range of BCI paradigms.

The structure consists of two main components that serve distinct roles in experimental workflows:

• MEDUSA Kernel. A Python library providing a suite of tools for advanced signal processing, machine learning, and deep learning, suitable for both offline analysis and real-time applications.
• MEDUSA Platform. A desktop application with a graphical user interface (GUI) for designing, executing, and managing BCI paradigms, aimed at simplifying experimental implementation and study management.

MEDUSA Kernel 

The MEDUSA© Kernel is the backbone of the ecosystem, providing researchers with powerful tools for brain signal processing and analysis.

Its functions are categorized into different levels of abstraction.

Medusa Kernel. Functions organized across multiple abstraction levels.

MEDUSA Platform

MEDUSA© Platform serves as the primary user interface. It has been designed with a modular, flexible and scalable architecture that can accommodate a variety of BCI and neuroscience experiments. Its structure is centred on three primary modules: 

Core modules and functionalities of the Medusa Platform.

Advantages for BCI Research

The platform MEDUSA offers a suite of significant advantages over existing platforms, making it a valuable tool for advancing brain-computer interface (BCI) research:

• Integration of advanced technologies. The software incorporates state-of-the-art methodologies, including deep learning models and functional connectivity metrics, into real-time and offline experimental workflows. This integration enhances the precision, depth, and versatility of data analysis, allowing researchers to extract more meaningful insights from neurophysiological signals.
• Support for novel BCI paradigms. MEDUSA provides built-in support for innovative BCI paradigms, such as c-VEPs, a highly efficient and reliable paradigm that is scarcely supported by alternative software solutions. This expands the scope of experimental possibilities, enabling researchers to explore cutting-edge paradigms with minimal setup complexity.
• Develop custom apps & experiments. MEDUSA provides a flexible environment for creating, deploying and sharing custom BCI applications and experimental paradigms. Its modular architecture allows researchers to tailor experiments to their specific needs. The platform includes built-in tools and templates to streamline the development process, reducing the technical burden on researchers.
• Tools for reproducibility and open science. The platform addresses critical challenges in reproducibility by offering robust tools for experiment design, customization, and sharing. Its integrated app marketplace and modular architecture promote collaboration and transparency, aligning with modern open-science principles and encouraging community-driven advancements.
• Ease of maintenance and scalability. Built on a Python-based framework, it facilitates straightforward maintenance and rapid prototyping of new features. Its modular and scalable design allows for seamless integration of emerging technologies, ensuring the platform remains relevant and adaptable to evolving research demands.

Applications and Use Cases

BCI Applications

Neurofeedback

Neurofeedback is a technique that provides the user with real-time information about their brain activity patterns. This biofeedback paradigm is interesting because, through operant conditioning, it allows the user to achieve voluntary control over the trained brain activity patterns. That is, to modulate them voluntarily. Thanks to brain plasticity, neurofeedback training can induce changes in the user's neural network (Sitaram et al., 2017). Therefore, this technique is gaining great interest as a possible non-pharmacological alternative for the treatment of pathologies characterized by abnormal brain activity. In this sense, neurofeedback protocols that seek to normalize pathological brain activity could be explored. 

MEDUSA offers an app for neurofeedback studies, allowing users to receive real-time feedback about their brain activity patterns. Among the main innovative features is its wide range of metrics to provide real-time feedback. In this sense, apart from the classic metrics based on power and power ratio, it also includes metrics based on connectivity. 

In addition, it provides different gamified neurofeedback scenarios to engage participants while training specific brain activities. For example:

• Scenario 1: A rising cube controlled by brain activity metrics. This scenario is designed for the user to have first contact with neurofeedback and learn how to voluntarily regulate target brain patterns.

Image displaying neurofeedback scenario 1

• Scenario 2: A spaceship whose vertical position depends on the value of the training metric. This scenario allows us to define a target change of the training metric (a percentage increase or decrease). In this way, the voluntary modulation of the target patterns is reinforced.

Image displaying neurofeedback scenario 2

• Scenario 3: A competitive race in which the user controls an avatar, which will move according to the value of the training metric. Like Scenario 2, this scenario allows to reinforce the training patterns by defining the modulation target during the trial. 

Image displaying neurofeedback scenario 3

The neurofeedback app integrated by MEDUSA is characterized by its configurability (Marcos-Martínez et al., 2023). In order to allow researchers to define their neurofeedback training protocols without limitations, this app allows them to define aspects such as the number of trials, the duration, the training metrics, the frequency band of the trained activity or the EEG channels used to calculate the feedback.

One of the most widely used applications of neurofeedback is in the field of neurorehabilitation of stroke survivors. The use of feedback on the activity of their sensorimotor region during the performance of imaginative motor tasks has shown promising results (Sebastián-Romagosa et al., 2020). Some examples of possible neurofeedback interventions that could use the MEDUSA app would be in alleviating the symptoms of attention deficit hyperactivity disorder (ADHD) or cognitive training of the elderly.

BCI Applications for Communication and Device Control

The traditional motivation behind BCI systems has been to improve the quality of life for individuals with disabilities by replacing hand-based control of applications and devices for everyday use. These applications are commonly referred to as “BCIs for communication and control” (Wolpaw & Wolpaw, 2012). 

In this domain, it is crucial to develop high-performance systems that enable users to select commands reliably and efficiently. To achieve this, such BCIs typically rely on exogenous signals that elicit real-time changes in EEG activity through visual, auditory, or somatosensory stimulation. These signals are then mapped to specific commands that users can select in real time, representing actions for an application or an external device. Among the control signals utilized in BCIs for communication and device control, the P300 evoked potentials, steady-state visual evoked potentials (SSVEP), and code-modulated visual evoked potentials (c-VEP) stand out.

P300 potentials were among the earliest control signals adopted during the initial decades of real-time BCI development by leveraging the traditional visual oddball paradigm. Numerous implementations of this paradigm have been described in the literature, with the most popular being the "row-column paradigm" (RCP). In this approach, commands are arranged in a matrix format, where rows and columns flash randomly. The user focuses attention on the desired command, generating a P300 evoked response in the parieto-occipital cortex approximately 300 ms after stimulus onset. 

By identifying the specific row and column eliciting these responses, the system can accurately determine the intended command (Martínez-Cagigal et al., 2019). Although P300 potentials are endogenous components, the system may also be considered exogenous because the entire event-related potential (ERP), including other visual-related exogenous deflections, is processed in the signal pipeline. These systems have consistently achieved accuracies exceeding 90% with information transfer rates (ITRs) of 10–25 bits per minute (bpm) for healthy users, and over 80% with similar ITRs for users with motor disabilities (Martínez-Cagigal et al., 2019). MEDUSA provides a free and customizable implementation of the traditional RCP paradigm. Users can tailor the system's run settings (e.g., number of sequences, colours, timing intervals, matrix design, etc.) and signal processing methods (e.g., rLDA, EEGNet, or EEG-Inception). More details are available at the Medusa website.

The performance of P300-based BCIs was quickly surpassed by systems utilizing SSVEPs or c-VEPs. Classical SSVEP systems operate by flickering each command at a specific frequency, generating an oscillatory response in the EEG that matches the frequency of the command being attended to by the user. 

These systems have demonstrated accuracies exceeding 90% and ITRs in the range of 40–50 bpm (Wolpaw & Wolpaw, 2012). In contrast, c-VEP systems encode command flashes using pseudorandom codes rather than constant repetitive tones (Martínez-Cagigal et al., 2021). One of the most common implementations of c-VEP systems is the circular shifting paradigm, where each command is encoded with a shifted version of a pseudorandom time series. This approach enables real-time gaze decoding by detecting phase differences in comparison to a pre-calibrated template. Because all commands are derived from variations of the same code, calibration requires recording the occipital visual-evoked response to the original code only. While both control signals exhibit generally comparable performance, c-VEPs offer several advantages. 

They are less susceptible to interference from narrowband baseline EEG activity (e.g., the alpha peak) and are less constrained by monitor refresh rates. Furthermore, recent research suggests that, from an information-theoretic perspective, the maximum achievable ITR through the visual-evoked pathway using c-VEP-based BCIs significantly exceeds that of SSVEP-based systems (Shi et al., 2024). MEDUSA provides two distinct, free-to-use applications that leverage c-VEP control: the “c-VEP Speller” and the “P-ary c-VEP Speller”.

While both applications share a common signal processing pipeline, they differ in their stimulation paradigms. The c-VEP Speller employs a classical binary black-and-white stimulus, whereas the P-ary c-VEP Speller introduces non-binary m-sequences that encode commands using varying shades of gray. This innovative approach has been shown to significantly reduce visual eyestrain while maintaining equivalent performance, achieving accuracies above 90% with information ITRs ranging from 40-100 bpm (Martínez-Cagigal et al., 2023).

A user wearing a Bitbrain Versatile EEG while controlling the P-ary c-VEP Speller.

Other Applications

Cognitive Neuroscience

MEDUSA integrates digital versions of classic neuropsychological assessment tests. In addition, these tests can be used in conjunction with EEG equipment. Therefore, the battery of cognitive tests included in the platform not only allows an in-depth and objective analysis of the user's responses but also to study changes in brain activity associated with cognitive processes (Marcos-Martínez et al., 2023). Consequently, these applications can be very useful in neuroscience research. For example, some of the tests could be used to investigate local and global changes in brain activity when a user performs tasks involving working memory. Similarly, these tests could allow the assessment of changes in cognitive functions associated with interventions such as Neurofeedback. 

The tests included in the Medusa cognitive assessment battery are: Dual N-Back, Digit span test, Corsi block-tapping test, Go/No-go task and Stroop test. All tests are highly configurable and include both English and Spanish versions. This allows researchers to define their study protocols without encountering limitations associated with test configuration. 

A user wearing the Bitbrain's EEG Diadem while conducting a Corsi Block Tapping Test.

Open Access Apps for Experimental Research

MEDUSA goes beyond cognitive psychology tests, neurofeedback, and BCI paradigms, offering a diverse range of free-to-use apps suitable for various experiments. For example, the Recorder app provides an intuitive framework for saving multimodal data synchronized with customizable events and conditions (e.g., eyes open and eyes closed). 

The Eye Artifact Recorder Paradigm app complements resting-state studies by enabling controlled recordings of ocular artifacts (e.g., horizontal and vertical eye movements, blinking), which are useful for calibrating artifact rejection algorithms. The PhotoMeasure app is designed for experiments measuring delays between screen-onset events and signal timestamp tagging, using tools like phototransistors. 

Additionally, MEDUSA offers applications tailored for educational purposes, such as the Visual Oddball Task Demo and the Checkerboard Reversal Demo, which provide real-time visualization of ERPs during oddball and checkerboard-reversal tasks, respectively. It is possible to explore the MEDUSA market to find a variety of innovative applications or contribute with your own designs to enrich the community.

Real Use Case with Bitbrain EEG Headset 

An online experiment was conducted with the c-VEP paradigm to evaluate the capabilities for BCI applications of Bitbrain’s dry EEG headset “Diadem". The results demonstrated exceptional suitability for practical BCI applications such as communication and device control. 

Bitbrain’s dry EEG cap “Diadem”

The device Diadem is equipped with 12 EEG dry sensors positioned in the pre-frontal, frontal, parietal, and occipital regions. During the experiment, the device exhibited robust and reliable performance, achieving high accuracy and high-speed control of MEDUSA's c-VEP speller. 

These findings underscore Diadem's ability to provide precise EEG signal acquisition with minimal setup complexity and prove Bitbrain’s dry-EEG system offers sufficient signal quality for BCI applications, meeting the demanding time and signal quality requirements typically associated with gel and water-based sensors.

Other Hardware Combinations 

In addition to the Diadem EEG headset, Bitbrain offers a range of EEG devices for human behaviour research and neurotechnology applications, featuring dry, water-based, and textile sensors.

From left to right, EEG headsets of Bitbrain include Water-Based Versatile, Dry-EEG Hero, and Textile-EEG Ikon:

• Versatile: A multi-channel EEG cap with 8, 16, 32 or 64 channels that uses tap water instead of gel, making it easy to use and set up. 
• Hero: A wearable headset with 9 dry sensors, it can be used for neurorehabilitation and motor imagery with the Medusa platform. Another possibility within the dry-EEG range of products is Air EEG, a compact system with 8 dry sensors, suitable for basic cognitive and emotional state estimation.
• Ikon: A textile-based EEG system designed for cognitive research and sleep studies. The Ikon Sleep version is specifically adapted for sleep monitoring.

These devices are designed to deliver accurate and reliable data while offering easy setup and user comfort, making them ideal for neuroscience research in the lab and real-world settings.

Software Integration 

One of the options to record data with these devices is SennsLite, a software tool that allows the user to run and export records from Bitbrain devices and synchronize them with others. 

Bitbrain's SennsLite Software 

Main features include the ability to record, and export synchronized data from multiple devices in CSV and EDF formats, offering flexibility for data analysis. It also supports Lab Streaming Layer (LSL), which enhances research by:

• Handling data with irregular sampling rates, essential for studies involving multiple biometric signals.
• Enabling real-time and offline data processing, improving study flexibility and accuracy.
• Providing easy access to synchronized multimodal data through its XDF file format, simplifying analysis and interpretation.

Integrating the Medusa platform with Bitbrain's EEG hardware and software presents advanced biosignal data collection and ensures synchronization and accuracy in multimodal studies through LSL and SennsLite.

Connecting MEDUSA and SennsLite through LSL Streaming

Bitbrain devices connect to MEDUSA through LSL using the SennsLite software. Here is a step-by-step guide to set it up:

1. Pair the Bitbrain device with your computer via Bluetooth.
2. Open SennsLite, go to the "Recording" mode and click "Add Sensor" if the device is not yet registered.
3. Configure the device settings, including signal type and dataset output preferences.
4. Enable LSL Streaming in the SennsLite configuration panel by toggling the LSL Streaming option for the device. This ensures that the device streams its signals to an LSL server once recording starts.
5. Launch MEDUSA and navigate to the LSL configuration panel. Bitbrain device’s stream should appear in the “Available LSL streams” tab.
6. Select the streams you want to integrate and click the button to add the desired streams to the workspace. Configure the acquisition in MEDUSA and click Ok. The stream should appear now in the “Working LSL streams” tab.
7. Configure signal visualization and start experiments in MEDUSA using the apps panel. 

This integration allows researchers to utilize MEDUSA with any Bitbrain device, including Versatile, Diadem, Air, Hero, Ikon, eye-tracking systems, and other biosensing devices. To achieve optimal results, it is essential to select the device or combination of devices best suited to the specific requirements of your experiment.

For detailed documentation and further instructions, refer to the following resources:

• Bitbrain Download Center 
• Medusa BCI Support

Conclusion

As an innovative platform, MEDUSA aims to drive advancements in BCI and neuroscience research by offering a scalable, accessible, and robust solution. Its goal is to continue supporting researchers in designing experiments, analyzing data, and exploring new dimensions of brain function and interaction, ultimately fostering progress in the field.

The vision is to expand the tool with advanced capabilities, including multi-platform support, enhanced real-time visualization tools, and novel experimental paradigms such as competitive and collaborative BCIs, tactile and auditory stimulation, and new neuropsychological assessment tools. 

Committed to open science and community-driven development, the platform encourages contributions from research groups worldwide, further enriching it as a growing resource for advancing and fostering innovation in neurotechnology and neuroscience research.

About the authors

Dr. Eduardo Santamaría-Vázquez, Dr. Víctor Martínez-Cagigal, and M.Sc. Diego Marcos-Martínez are leading researchers in brain-computer interfaces and neurotechnology, in addition to their contributions to the development of the MEDUSA software ecosystem.

• Dr. Eduardo Santamaría-Vázquez (Scholar, Research Gate, Linkedin)
• Dr. Víctor Martínez-Cagigal (Scholar, Research Gate, Linkedin)
• M.Sc. Diego Marcos-Martínez (Scholar, Research Gate, Linkedin)

You might be interested in

• What is BCI? An introduction to brain-computer interface using EEG signals
• Bitbrain Bringing Brain Computer Interfaces BCIs to the user’s home for neurorehabilitation and assistive applications
• EEG test: Uses, Procedures and Risks
• What is EEG and what is it used for?
• Future and ethics of neurotechnology and brain-computer interfaces
• Modern BCI-based Neurofeedback or EEG Biofeedback for Cognitive Enhancement

References

Santamaría-Vázquez, E., Martínez-Cagigal, V., Marcos-Martínez, D., Rodríguez-González, V., Pérez-Velasco, S., Moreno-Calderón, S., & Hornero, R. (2023). MEDUSA©: A novel Python-based software ecosystem to accelerate brain-computer interface and cognitive neuroscience research. Computer methods and programs in biomedicine, 230, 107357.

Martínez-Cagigal, V., Thielen, J., Santamaria-Vazquez, E., Pérez-Velasco, S., Desain, P., & Hornero, R. (2021). Brain-computer interfaces based on code-modulated visual evoked potentials (c-VEP): a literature review. Journal of Neural Engineering, 18(6), 061002.

Martínez-Cagigal, V., Santamaría-Vázquez, E., Pérez-Velasco, S., Marcos-Martínez, D., Moreno-Calderón, S., & Hornero, R. (2023). Non-binary m-sequences for more comfortable brain-computer interfaces based on c-VEPs. Expert Systems with Applications, 232, 120815.

Martínez-Cagigal, V., Santamaría-Vázquez, E., Gomez-Pilar, J., & Hornero, R. (2019). Towards an accessible use of smartphone-based social networks through brain-computer interfaces. Expert Systems with Applications, 120, 155-166.

Shi, N., Miao, Y., Huang, C., Li, X., Song, Y., Chen, X., Wang Y., & Gao, X. (2024). Estimating and approaching the maximum information rate of noninvasive visual brain-computer interface. Neuroimage, 289, 120548.

Wolpaw, J. R., & Wolpaw, E. W. (2012). Brain-computer interfaces: principles and practice. Brain-computer interfaces: principles and practice. Oxford University Pressdoi.org/10.1093/acprof:oso/9780195388855.001.0001

Marcos-Martínez, D., Santamaría-Vázquez, E., Martínez-Cagigal, V., Pérez-Velasco, S., Rodríguez-González, V., Martín-Fernández, A., Moreno-Calderón, S., Hornero, R. (2023). ITACA: An open-source framework for Neurofeedback based on Brain-Computer Interfaces. Computers in Biology and Medicine 160, 107011.

Sitaram, R., Ros, T., Stoeckel, L. et al. (2017). Closed-loop brain training: the science of neurofeedback. Nat Rev Neurosci 18, 86–100. 

Sebastián-Romagosa, M., Cho, W., Ortner, R., Murovec, N., Von Oertzen, T., Kamada, K., Allison, B., Guger, C., (2020). Brain-Computer Interface Treatment for Motor Rehabilitation of Upper Extremity of Stroke Patients—A Feasibility Study. Frontiers in Neuroscience, 14, 591435.