 My name is Timur Dixite. I'm working in in Yulich with Cartridge Ammuz. I'm a computer scientist, not a neuronatomist, so the expertise is overlapping but highly complementary here. I try to give you a first impression of what you can do with the human brain project tools with respect to the atlases. But just to quickly summarize again, so the way that we look at an atlas is usually on the high level. We have a definition of a coordinate space, which usually comes with an image of a brain in that coordinate space. Then we have a map, so a delineation of the coordinate space into different regions, which might be more complex than just having clear borders, but it's in principle it is a map. And then we have a terminology or a taxonomy on top of that map. So basically something that tells us what are the names of these different elements and how do they relate to each other. It might be hierarchy or whatever. So these are the three basic parts that Drikwe has already mentioned. And then in each coordinate space we might have different maps for sure. That's not different from how we are dealing with maps of the Earth, where you can have a political postulation or something regarding climate regions or the topography and things like that. We also want to have a flat version of your map too, right? Yes, that is also an analogy that's true. So for the Earth we usually look at a flattened version and that is what some people do for the brain as well. Oh yes. And then in the HEP we currently support a range of coordinate spaces. Drikwe already said it. It's the L-Miles, the Waxhorn Red and different reference spaces for the human brain. Before we look into the tools let me just give you a little idea of the postulations that we use in the human brain because it's a bit different. Maybe for some of you that is something well known, for others maybe not. So in the human brain we have this particular, maybe I just turn the slides around, we have this particular situation that the brain is highly folded and highly different in different individuals. So if you take brains of ten different humans and you map in this case the visual areas, the primary and secondary visual areas, you would see that the location of these areas is very, very different in the different brains. So here we have this problem that it is very, very difficult to give a good explanation in the reference space because it might be very different in the individual subjects. And that is something we do in really for many decades and which is the basis for the Cytoarchitectonic Probabilistic Atlas that we support in the human brain project. So here we typically have one scientist or often it's a PhD student delineating a single area in the human brain and they do it in histological sections where we, at the one micron resolution, where we actually try to find the differences in the laminar patterns of the cell distributions. And what they will do is they take histological like few hundred or less than hundred histological sections in one brain, try to find these borders and then there is an algorithm that can confirm, a computer program that can confirm that there is really a significant border so it's not arbitrary and they will do it in many sections in one brain, but then in ten different brains so that we really capture a little bit of the viability in the human brain. And then these maps will, these 2D delineations are projected into a three-dimensional, common three-dimensional space where we can then compute a probability map which basically shows us, as you can see here on the right, in red would basically mean these voxels have been assigned to this area in ten of ten brains. And if you see here blue colors or low probability it means at this position we have delineated this area only in maybe in one of these ten brains. So it captures a coarse estimate of the probability of the viability that this area has in this particular brain and then we provide these probability maps in the reference space. They're currently provided in two reference spaces in the MNI column 27 space. This is a single subject space where you really see very precise the folding patterns, the gyri and salchai of this brain and also in the MNI 152 space which is an average across multiple subjects. The second human template that we provide is the so-called big brain. This is a very different thing, it's a single subject template and here we have actually taken all the histological sections of one particular human brain which is 7400 sections and reconstructed them back into a consistent 3D representation at a resolution of 20 micron so at a very high resolution. So this is a template space where you have only one brain but you can really zoom into 20 micron resolution and see the different laminar layers in the brain so it has a very high resolution. And the strategy is that we provide in the big brain we provide also maps of the same side to architechronic areas but referring to this one particular brain so you have a link from the high resolution maps in this microscopic space to the probabilistic maps in the millimeter resolution which capture the viability so you have a choice to basically see the viability or to go into very high detail for this one single subject. And as you will see later a strategy is that we actually collect and anchor multimodal data that has a high resolution and a high spatial resolution in this space so if we really for anchoring data if we really need the resolution that distinguishes different cortical layers and so on we will use this space as a reference space and if we have signals at the whole brain level or very low resolution signals it can be done at the millimeter resolution with the probabilistic atlas and so we have a link between the two. So that one has also MRI so she would save exact brain and everything? There is an MRI for this so this has been published in 2013 in this brain so this is an MRI but the MRI is not of very good quality so in the atlas we don't have the MRI of this volume it's really a 3D representation of this histology reconstruction but this is not the big brain it's not a product of the HPP it has been done between Ulish and the lab of Allen Evans in Montreal and it's available free for download. The way we support it here is that you have it in the viewer it's anchoring of lots of data that we provide through the human brain project in this space and there are low resolution versions at the MRI scale so there is actually a link to the lower resolution scale so you can enter it through the MRI resolution level but there is no high quality MRI available for this one the quality is not so good so this is actually already a picture of the viewer that we have they zoomed into the big brain and you can see that this goes close to the resolution of single cells not exactly you do not see the morphology of single cells but it comes close and with this big brain we provide a first set of cytokinetic maps in 3D some of them are very high detail they have been computed using machine learning algorithms with supervision some of them are interpolated and not so high detail but this is now iteratively filled up we also have maps of the cortical layers as you can see on the right and if you look at this in the browser we will do it later on you can really see it's you really have a resolution in the cortex in the different layers and can really nicely see the architecture here is a map of the cytokinetic areas and you can see nicely here this is the border between the primary and secondary visual cortex you can see how the arrangement of the cells and the layers is changing here and the 3D map is actually confirming this that's where you find the map so then we have this online viewer which develops piece by piece and it allows to browse all these brain templates it's just a little move and you can see how it allows to zoom into the high resolution in the big brain you can do oblique slicing so you can get arbitrary angle sections you can select the different passillations you can choose another template this is the probabilistic cytokinetic atlas and then you can browse the region hierarchy, find brain regions if you double click them it takes you to the region and it allows you then from there also to search for data that is linked to this area so in this case you see here in the movie you can find samples of receptor density measurements in this area just by visually browsing the 3D volume and looking at the areas this is something we will do together during the day there's a map of fiber bundles being developed and here's just the same thing in slow so basically when you enter an atlas you find the region hierarchy on the top right and you can type and start typing the name of an area it will reduce to all the text matches of this area so it's quite easy to navigate double click you get there and then if you right click it is doing a query to the database of data sets that we have in the human brain project which have been linked to this area and this actually relies on the work of the curation teams Camilla has outlined so you wouldn't find anything if there wouldn't be a curation team basically making sure that there's the metadata that links this data set to this region and then from there you can actually like in this example for example I have again the primary visual area I will see here that there's also a map of that area in the big planet high resolution and I can click on it it will take me to the knowledge graph which is the database with all the data sets we will also look at that data with all the information about this data set so there's a long description, there's a DOI, there's a publication you find a license under which you can use that data set you find the files and related publications and you also find the link that takes you into the atlas viewer so it's cross connected, you get from the viewer you can go into this database with all the rich information and back in this case if I go back because this data set is a map that has been mapped in the high resolution big brain space I would automatically get back into the data space so I would now see the same area mapped in the big brain and just to conclude before we play with it so in principle, browsing atlases in the viewer is the same thing as searching data sets in the text search in the human brain project similar as you would have it in Google so you might search for restaurants in Warsaw and you will get a text list of these restaurants but you can also switch to Google Maps and you will get that same restaurants in the area pinned to these locations so it's the same effect but let's get our hands on it and the idea was that we start with the rodent plane so I think there were some instructions distributed before the course already that you should have a recent version of Firefox or Chrome on your computer to run these exercises so if you maybe tricky you want to take over at that point so the idea is that we use now the Atlas viewer to look up the rock brain atlas and to basically explore the hippocampus I don't know how many of you are well known with the anatomy of the hippocampus or able to find the boundaries of the subregions there but the idea was to say well the hippocampus is a very distinct structure in the brain it has this sausage shape, it's curved and it is considered to be quite complex so it has many different layers in subregions that can be difficult to identify and the use of an atlas is very important to find that properly and of course the hippocampus can also be challenging to navigate it if you have orientations of sections or images that is not standard the shape in the human resembles of course the sea horse hence the name hippocampus you also have the corneum among the CA fields resembling the curved shape of the almond's horn so I think that if you move to the next slide people the task was to go to the HPP viewer there is a link here where you can also enter it through the human brain project website and then go to explore atlases launch the viewer that Timo just demonstrated and then select the voxel space rat brain atlas with the lineations try to use this to simply find the ventate gyrus and it's bound with the CA3 field and then try to use the potential of the viewer to cut through this region in 3D to just look at how its shape changes as you change the angle of the viewer and then after that we will have a similar type of exercise with the human brain just try it out maybe quick instructions so you move the images by simply clicking and dragging and if you hold shift and then click and drag you will get oblique slices so you will basically rotate by shift click and drag and zooming is simply with a scroll wheel it should be intuitive and there is also the big brain available and we just suggest that you try to find the primary motor cortex in the human as a second exercise and if you find it in the big brain it is interesting to see if you can spot some of the giant bit cells which you have in the motor cortex it takes a moment and in the search list you should among other entries find the probabilistic maps so here in that case is the probabilistic map and you see a little eye which allows you actually to see the probabilistic map so if I hit this little eye it will show me the probabilistic map and it will at that moment remove the rest of the atlas so you see a huge difference this is really the probabilistic map and if I disable it again you see how the same area in the maximum probability map is much more restricted so the real rich information is in the probability maps and I just suggest this by right clicking and searching for related data sets and if you are interested to see what this probabilistic map is about you can just click on the name of this data set and it takes you to the knowledge graph page with the descriptions and then for the second task so if you just select here the big brain and you just zoom in with the scroll wheel or on a Mac touchpad with two fingers it just takes you to the high resolution actually I think I'm already here in the motor cortex