Browsing the Streamline Database¶
The goal of this workflow is to load some data into the DSI2 Browser and be able to interactively query and cluster streamlines.
Assuming that you have imported your data through dsi2_import, you can launch the data browser from your shell:
$ dsi2_browse
Red Text: Data Source and Aggregator¶
Launching the Browser, you can select which individual datasets will be queried when you launch the Sphere Browser.
Select your Data Source in the top left of the screen¶
Will you be loading data from your local hard drive or querying a remote database? You can choose either of these options from the Data Source listbox.
Change Aggregation Algorithm to Region Labels¶
Select which streamline aggregation method you will use to label the streamlines returned by your searches in the Voxel Browser. In this example we will choose a Region Label Aggregator, which groups streamlines based on which regions they connect. Other aggregation options include k-means clustering or DiPy’s QuickBundles algorithm.
Yellow Box: Specifying data source properties¶
- Choosing a data set
Select your -.json file
- Enter either a Scan id, Subject id, Study, or Scan group to search for an individual dataset
- These fields can be used to search for specific properties of datasets.
- Once your specifications are filled out, click Search for Datasets highlighted in green above
- This will populate the list of results
Blue Box: Matching datasets¶
Datasets matching your search criteria are listed in the blue box above.
The first few columns present information about the individual who was scanned. The rest offer options on how the streamlines from each dataset will be displayed in the Sphere Browser.
If the dynamic color clusters column is checked, then the colormap from the color map column will be applied to that dataset’s streamlines. This is useful for interactive clustering in the Sphere Browser since you can see which group each streamline was assigned to based on its color.
On the other hand, if you are interested in plotting streamlines that are colored only according to which individual they came from, then dynamic color clusters should be unchecked and a static color can be assigned.
If there are datasets in the list that you would not like to be included in your Sphere Browser, click anywhere in the row and click the delete row button.
Once you are happy with the list of datasets, click the **Launch sphere browser* button in the purple box.
The Sphere Browser: Querying and visualizing streamlines¶
The Sphere Browser lets a user choose a set of coordinates, queries the data source at those coordinates, then aggregates the streamlines before displaying them in the 3D viewer.
Selecting Coordinates: Using a sphere¶
A sphere is a handy way to select a set of coordinates.
Sliders in the green box let you move the sphere in x, y and z coordinates and increase its radius. Moving these will directly affect the visualization of your search coordinates, which appear as red cubes (circled in the 3D viewer in orange).
Selecting Coordinates: From an ROI¶
If there is a set of voxels defined in a NIfTI file, you can load them into the browser by selecting Data ‣ Search from NIfTI. You will see the following dialog box:
Select a .nii file in the Filepath box. If there are multiple regions in the file you’d like to include (for example a region labeled 1 and a separate region labeled 2) you can right click the arrow to include a second region. If you’d like to expand the regions in space, you can dilate them by n voxels. Clicking OK will clear any previous search coordinates and render your new coordinates in the 3D viewer.
Visualizing Streamlines¶
Widgets in the magenta box provide control over how streamlines are plotted. If Auto Aggregate is selected, the cluster assignments will be updated any time one of the aggregation-related widgets is interacted with. In the case of this image the aggregator is a k-means aggregator, which has two parameters. k, or the number of clusters to define, and min_tracks as the minimum number of streamlines that must be assigned to a cluster before the cluster can be considered legitimate. As these two sliders are dragged around, the new cluster assignments are visible in the blue box. The color of each row in this list corresponds to the color of the streamlines in the 3D viewer.
The Render tracks checkbox can be used to prevent the rendering of stremalines in the 3D viewer. If Auto aggregate is enabled while Render tracks is disabled, cluster assignments will still be visible in the cluster list and will be updated as you interact with the clustering widgets.
The Render clusters checkbox used to render glyphs that summarize each cluster. This functionality no longer works.