The tool, designed by University of Granadais called MVPAlab and is capable of performing detailed pattern analysis on data from brain electroencephalography.
A multidisciplinary scientific team from the University of Granada (UGR) has designed the application MVPAlabwhich thanks to programming and artificial intelligence (AI) offers an intuitive and easy-to-use tool for the analysis of electroencephalograms of the brainexplorations that are key in the human neuroscience. This work exemplifies the existing collaboration between university institutes of the UGR that unite the AI with the study of the human brain.
The group of researchers leading this advance is made up of David López and José MG Peñalver, of the Mind, Brain & Behavior Research Center (CIMCYC), Juan Manuel Górriz Sáez, from the Interuniversity Institute of Data Science and Computational Intelligencehe (DaSCI), and María Ruz Cámara, director of CIMCYC.
This tool has an igraphical user interface that allows you to configure and launch different types of brain scans, as well as graphically represent the results obtained in a visually appealing way, all without writing a single line of code.
MVPAlab implements machine learning-based algorithms that perform multivariate analysis of brain patterns, cross-classification, temporal generalization, and analysis of the contribution of each of the electrodes and frequency bands to the final classification result.
In addition, the application has a set of preprocessing solutions for, among others, the normalization, balancing and reduction of the dimensions of the data obtained, the smoothing of the signal and the reduction of the signal-to-noise ratio. At a statistical level, MVPAlab performs a non-parametric permutation analysis, based on clusters, which makes it possible to obtain statistically significant regions in the brain scan, in order to make inferences at the group level.
“The study of brain function using electroencephalography has relied for years on univariate methods. The development of science and technology in recent decades has fostered the appearance of new and more complex techniques, based on statistics and artificial intelligence, which allow data to be explored in greater depth.”, details the study associated with the implementation of MVPAlab.
The research team of the UGR who works on the application considers that, despite the tremendous effort put into creating tools that bring these methodologies closer to various areas of science, today “Its use and implementation is still complicated, especially in laboratories with little programming knowledge”. To future, MVPAlab it evolves towards its constant development, implementing and improving functionalities continuously. One of the lines of work is the implementation of fusion techniques multimodal.
These techniques make it possible to jointly analyze data from different sources. neuroimaging No invasive (such as electroencephalography and functional magnetic resonance), combining their strengths and minimizing their shortcomings. The use of these solutions represents a great step forward in the study and understanding of brain function.. The source code of MVPAlab It is publicly hosted in a GitHub repository (under the GPL v.3.0 license, which allows users to use, modify or share this tool freely.
In this same link you can find the complete documentation of the tool, different tutorials, data sets and test scripts, as well as a discussion forum where users can suggest new features or report and track errors.
