MEDUSA© Kernel

Tutorials

Would you like to learn how the MEDUSA-Kernel works? Don't miss our interactive tutorials!

For MEDUSA© Kernel 1.3.0

If you have any questions that are beyond the scope of this help file, please feel free to ask for help in the forum or contact with us.

The MEDUSA© Kernel is a powerful tool for analyzing brain signals, offering advanced capabilities in signal processing, machine learning, deep learning, and high-level analyses. It's also utilized by the MEDUSA© Platform to handle multiple biosignals, save experimental data, and implement paradigm-dependent signal processing pipelines. We understand that the framework can be complex, and the learning curve might be challenging. To help you get started, we've prepared a set of tutorials. Check them out below!

Algorithms

The Algorithm class is a powerful tool that provides common ground for data processing pipelines as well as persistence functionalities. If you need to define and distribute standalone algorithms with full compatibility with MEDUSA native methods as well as third party packages, this is your tutorial. The notebook will cover the main features, functions and classes involved in the definition of an algorithm through illustrative examples. In this notebook you will learn:

  • What is the ProcessingMethod class
  • Wrap functions and external classes in the ProcessingMethod class
  • Define a processing pipeline
  • Create an algorithm
Access the Tutorial Here

c-VEP Analysis

Code-modulated visual evoked potentials (c-VEP) are prominent in the non-invasive BCI literature for their ability to enable reliable, high-speed control. These control signals arise from the brain's exogenous responses to flickering stimuli that follow pseudorandom codes. In this notebook you will learn:

  • Load electroencephalographic (EEG) signals from a c-VEP experiment
  • Plot the pseudorandom sequences that encode application commands
  • Calibrate the system using the reference processing pipeline
  • Plot the calibrated templates
  • Decode the selected commands during a free-spelling phase and evaluate performance metrics
Access the Tutorial Here

ERP Spellers Module

This module contains high level classes and functions specifically designed for spellers based on event-related potetials (ERPs). This notebook will cover the main features, functions and classes of the module through illustrative examples which will show you the power of the included tools. In this notebook you will learn:

  • What is an ERP-based speller
  • Download an open ERP-speller dataset and explore the files
  • Create an instance of ERPSpellerDataset
  • Know the feature extraction and decoding functions included in the module
  • Implement an asynchronous ERP-based speller using the built-in models
Access the Tutorial Here

ERP Models

MEDUSA is a framework designed for scientists and developers who investigate novel signal processing algorithms, reducing the development and testing time in real experiments. This includes not only the implementation of cutting-edge signal processing methods, but also high level functionalities to assure the persistence and reproducibility of the algorithms created within the framework. One of they key features that makes MEDUSA so powerful is its ability to implement and share standalone algorithms out of the box compatible with Medusa applications. In this notebook you will learn:

  • How to create a custom model for ERP-based spellers
  • Save the algorithm
  • Use the algorithm in MEDUSA© Platform
Before this tutorial, make sure you have checked: Access the Tutorial Here

Local Activation

MEDUSA's local activation module implements several parameters focused on analyzing the signal of each sensor individually. That is, the parameters in this category return a single value for sensor and epoch. The parameters here can be divided in two sub-categories: spectral and non-linear. The spectral parameters measure different properties of the signal's power spectral density. On the other hand, the non-linear parameters assess the time courses of the signals, evaluating its non-linear properties such as the complexity, irregularity, or variability. This notebook will cover the main functions of the module through illustrative examples that will allow you to get started in biosignal analysis with MEDUSA. In this notebook you will learn:

  • Download an open EEG dataset and explore the files
  • Pre-process the EEG signal with MEDUSA filters
  • Know the different spectral parameters
  • Know the different non-linear parameters
  • Get some plots with MEDUSA's plot functions
Access the Tutorial Here

Connectivity

MEDUSA's connectivity module implements several methods focused on analyzing the relationship between the signals recorded at different sensors. This methodology provides a global view of the relationships between brain activity in different regions. In this sense, the different methods implemented in MEDUSA return adjacency matrices, i.e., matrices in which the off-diagonal elements account for the connectivity value between two sensors or two signal sources. Within the MEDUSA's connectivity module we can find two sub-categories: connectivity based on signal phase, and connectivity based on signal amplitude. This notebook will cover the main functions of the module through illustrative examples that will allow you to get started in biosignal analysis with MEDUSA. In this notebook you will learn:

  • Download an open EEG dataset and explore the files
  • Pre-process the EEG signal with MEDUSA filters
  • Know the different phase-based connectivity methods
  • Know the different amplitude-based connectivity methods
  • Get some plots with MEDUSA's plot functions
Access the Tutorial Here