BrainX

A tool for Visualizing EEG Data

The aim of this project was to design a visualization system to visualize EEG data using seizure data as a test case.

To download the full project and see my code, go to my github page here.

Project Overview

Problem

EEG data is usually displayed as a standard line graph, however since EEG data is complex it is possible that a researcher could miss something important in the data as there are many overlapping lines in an EEG line graph.

Solution

By making the EEG data 3D and projecting it accurately onto a 3D model of a brain researchers may be able to pinpoint what is going on in certain regions of the brain better. BrainX is a 3D visualization tool meant to assists researchers in the investigation of brain data and it meant to be supplementary to typical line graphs.

Design Process

Research Stage

This project was done in conjunction with SPECS lab in Barcelona, a neuroscience & technology research lab. The first step of this project consisted of interviews with the researchers at SPECS lab asking them what their issues and needs are when working with EEG data. From these interviews a need for a visualization that allowed researchers to view EEG data in 3D space came up.

Design Stage

Sketching & Storyboarding

For the first step of the deign stage I did some sketching and then some storyboards for brain activity could be visually represented. The lab I was working with provided several csv files of simulated brain data, and from the previous interviews I knew that there was a desire for a 3D visualization. Since I was given time series data it was necessary to have a UI element that allows the user to view different points in time, leading to the inclusion of a timeline element.

I needed a way to indicate the EEG signal strength, as an abnormally high signal indicates a seizure. For this I used two visual indications, heatmap coloring, and the height of EEG spikes. Yellow and orange coloration indicates abnormally high EEG activity, as do longer EEG spikes, where as green and blue coloration and shorter spikes indicate normal brain activity. The heatmap allows users to quickly spot abnormal activity, representing seizures with longer spikes lets the viewer spot a potential seizure better in a 3D environment, as they can see the longer spikes from the other side of the brain model. There were 2 levels of seizure strength, light seizure and heavy seizures. Light seizures had fewer or even no orange spikes while heavy seizers had more.

Interactive Prototype

In the interactive prototype, the users have a few ways that they can scrub through the timeline and view the brain model. To rotate around the brain model independently a user can click the left mouse button and drag around the model, the model will then rotate in the direction that the user drags the mouse, and they can zoom by using the scroll wheel. The user can click the left and right arrow keys to rapidly scrub through the timeline at the top left corner

Alternatively the user can click and drag with the left mouse button which will scrub through the timeline and will also rotate the model gradually along the x axis, showing them both the changes over time and a full view of the brain simultaneously.

Evaluation Stage

Usability Testing

In order to evaluate the BrainX tool a group of 28 grad students were recruited to participate in a study, all of the grad students were familiar with EEG data from their own work. There were 20 samples of simulated EEG signals which the participants were asked to look at, each participant was asked to view the same EEG sample twice, once in BrainX and once in EEGlab, for a period of one minute per system. EEGlab is a popular 2D line graph EEG visualization tool so it was a an ideal choice for comparing BrainX to as our research goal was to see if finding and identifying a seizure was easier when viewing a 3D model versus a 2D graph. Whether participants viewed a given a sample in EEGlab or BrainX first was randomized.

There were 4 types of sample EEG signals that participants viewed, general seizures right temporal lobe seizures, left temporal lobe seizure, and healthy brain signals, all of the sample EEG signals were 10 seconds long. The participant’s were asked to identify which type of sample EEG signal they were looking at based on the 4 types, and if there was a seizure at what second did it begin at? The participants were not told what indicates a seizure in the BrainX system and were left to intuit that on their own as we wanted to be sure that how we chose to represent a seizure was recognizable. Ultimately we did no find any significant difference between how participants performed in BrainX versus EEGlab.