As this week’s topic was brain connectivity, one may find it an interesting question how different types of connectivity information can and should be displayed to the user. This tries to be a brief overview of some methods that have been used in the last decade or so. As we already noticed in the very first exercises of this course, brain anatomy is difficult. It is as if the brain was more complicated than all the other body parts and organs summed up. When you see anatomical sections for the first time, it is (at least for many people) immensely difficult to visualize the information in 3D. If one manages to imagine the brain in 3D, it is far more difficult to imagine brain connections – countless of them – in 3D, based on some connectivity matrices or some format of 2D illustrations. Luckily, today’s technologies are here to rescue us from this problem. Despite being very useful overall, playdough does not help much with visualizing connections, after all!
Historically, it has been difficult to make sense out of the brain. As we know, the brain surface has countless folds – sulci and gyri. Sometimes, the brain is presented as though having a smooth surface. Still, drawing a smooth or smooth-ish brain surface gives a false impression about the proximity of two points on the surface. Furthermore, colors are sometimes used to display the folds. However, if other information needs to be displayed, it would be wiser to save colors for that purpose. Fortunately, we now have the tools to build quite accurate 3D models of the brain. These models can then easily be rotated and zoomed into. They can enable changing the opacity of some layers to see others, as well as opening up the folds of the brain to a wanted degree. So, who builds these models and how? How accurate are they?
One huge project is the Human Connectome Project, also mentioned on the lecture. In this project, analysis is performed e.g. on the time series data of brain function and functional connectivity maps are formed. On the other hand, The Allan Institute for Brain Science collects data in their huge, free database: The Allen Brain Atlas, in which they started a large-scale program to gather data about the human brain in May 2010. The Atlas maps gene expression across the human brain, combining genetic and anatomical information. They slice donated, frozen brains into thin micro-slices and, for example, mark RNA molecules with in situ hybridization techniques. Thus, they acquire information of which genes are expressed and to what degree. Combining this with anatomical information acquired with MRI and DTI, they are able to produce highly detailed 3D models which can be viewed with their Brain Explorer 3D software. With this tool, researchers can investigate clues to the role and function of genes of interest in disease. So, this was already great back in 2013, but what is done now, four years later?
Screenshot of the Allen Brain Atlas database
Today, we have several technologies which utilize the idea of gamification in mapping brain connectivity. One of these is Mozak – the brainbuilder, and another is Eyewire. Mozak attempts to trace and classify neurons from neuronal 3D images, whose structures are branched and thin and therefore difficult to correctly detect with computational methods. Humans can help in this by playing the game. Eyewire – like its name suggests – aims at discovering how neurons process visual information. At the same time, the user-generated reconstructions can be used to train their AI further.
The explosion in computational power and novel technologies are beginning to also enable real-time analysis of brain circuits. Hopefully, what is utilized in research, will soon be available for educational purposes, and we can enjoy the 3D maps on our lectures and exercises. The VR hype is widespread, so this is to be expected. One of the TEDx Talks presenters compared the development of brain visualization techniques to the birth of Google Maps – and not without a reason.