DUGi

pyVizfMRI : A tool to visualize fMRI Signals Master’s Thesis in Data Sciences

http://dugi.udg.edu/item/http:@@@@hdl.handle.net@@10256@@27575

This Master Thesis aimed to find new ways to visualise fMRI signals and to provide a tool to help researchers to visualize and investigate Functional Magnetic Resonance Imaging (fMRI), Electroencephalogram (EEG), and Magnetoencephalography (MEG). The tool will incorporate the most used Observables of the library developed for the European project BRICON HBP SGA3 of the European Union’s Horizon 2020 Framework Programme. In the literature, there are several tools designed to help researchers in this field. However, none of them use the library mentioned before and none of them offer the combination of features provided by our project, like compatibility, user-friendliness, domain relevance, active support, and adherence to open-source principles. The brain data is represented as 2D arrays of signals over time. The data was sourced from the Human Connectome Project due to its accessibility and compliance with US regulations. Our tool is called "pyVizfMRI". Addressing data loading challenges, plugins and a FileConverter class enable custom converters for diverse file formats. The implementation includes an Observables gallery for visualization, brain mapping with various parcellations, and additional features like saving charts, modifying charts, rearranging BOLD signal ranges, customizable layouts, and multi-signal analysis, enhancing user experience. We also discuss the challenges faced by current brain signal visualizations. It highlights the limitations of existing methods, such as oversimplification and loss of information, due to reliance on techniques like heatmaps. Different visualization techniques are explored, including heatmaps, chord diagrams, network diagrams, arc diagrams, and edge bundling, each with its pros and cons. We propose a new visualization approach that combines elements from arc and chord diagrams to address the shortcomings of current methods. The implementation and testing of the pyVizfMRI framework are presented, showcasing its effectiveness in visualizing and analyzing brain data. The tool offers versatile and user-friendly features, including timeseries visualizations, functional connectivity heatmaps, sliding window analyses, phase-based analyses, and graph-based connectivity visualizations. Additionally, a 3D brain mapping visualization was developed for better interpretability. The intuitive GUI enhances user experience with customizable options. Finally, a novel visualization method is proposed to improve the representation of Functional Connectivity data, offering a variation of a chord diagram with a dual-ring structure that enhances clarity and interpretability over traditional visualizations.

3

9