Visualization

You can visualize and explore the segmentation and image data using the web based visualization tool Neuroglancer developed with the Connectomics at Google team by Jeremy Maitin-Shepard and extended by members of our collaboration with some additional features. This page will give you a basic introduction to using Neuroglancer.

Basic Navigation

Main 3D controls:

Click and drag to rotate

Control+Scroll to zoom in and out

Shift Click and Drag to pan

The basic neuroglancer window has panels which reflect either 2d cross sections through the data, or a 3d visualization of selected meshes of reconstructed neurons. At the top of the window are tabs that reflect different layers, such as the “img” [Image] and “seg” (Segmentation) layers shown in the video above. These layers can pull data from different sources into your viewer.

Segmentation layers visualize the distinct objects in the layer as random colors. Pressing “L” will randomly change the color of those objects. You can find all the helpful shortcut keys by clicking the “?” icon in the upper right. Double clicking on a 2d segmentation layer will “select” the object and start visualizing its mesh in the 3d window. If you right click on any point in 2d or 3d, this will recenter the viewer around that point. Scrolling when in a 2d window will change the slice in the dimension not shown, so if you scroll in an XY view, you’ll change Z. You can also zoom in and out by either holding ctrl+using the mouse wheel, or pinching/expanding on a multi-touch bar. Zooming is linked across 2d views, but you can independently zoom the 3d view. Pressing space bar when highlighting one of the panels will change the view to only showing that panel. Pressing it again with a single panel shown will bring up all 4 view panels (XY, YZ, XZ, and 3D). Pressing shift+space bar while hovering over one panel will bring up that panel plus the 3D mode. You can find a larger set of keys and controls by pressing the ? mark button in the upper right.

If you right click on the segmentation layer you will bring up more information and controls about the layer on the right hand inside. Here you can control the opacity of the selected and unselected objects, as well the transparency of objects in 3d. You can also see the object IDs you have presently selected, change their colors individually, deselect them by clicking on their ID, or temporarily remove their visualization by unchecking the box next to them. There are also buttons for copying all the IDs you have selected, or only the IDs that are presently visualized to your clipboard. You can also paste in lists of IDs separated by a comma in order to select those IDS. This can be a useful way to move back and forth between neuroglancer and programatic queries.

Annotations

Users can add annotations of different types to the dataset using an Annotation layer. To add an annotation layer to the viewer, hold ctrl and press the + button next to the list of layers. Right click on the annotation layer to bring up the annotation layer control in the right hand panel. There you can select different annotation tools with the buttons located next to the color selector. Hover over each tool for a description of what it does. You can add annotations generally by selecting an annotation tool, then hold ctrl while left clicking in a panel (either 2d or 3d). Some tools require multiple clicks such as two clicks to set either end of a line, or set the center of the ball and the second click controls how large the sphere is. Our deployment of Neuroglancer contains a few more annotation tools than are presently available in Neuroglancer, as well as buttons to make downloading a csv or importing a csv of annotation easy.

Link Sharing

One of Neuroglancer’s key features is that everything about the state of the viewer is encoded in a dictionary state. Whenever you zoom in and out, select or deselect an object, add an annotation, or change just about anything, that state is modified. You can then package up that state and send it to a colleague by pressing the share button in the upper right. You can then copy the RAW_URL to share with your colleagues. Note if you are already familiar with Neuroglancer, in our version of Neuroglancer, the URL doesn’t contain the state information, but instead contains this local_id, which will work to reload the state on the computer that produced it, but not on other computers. This alleviates a performance issue we were seeing with modern browsers and rapidly changing large URLs.