manual-segmentation.md

Manually segment in standard space

To manually segment brain regions and tracks in standard space that can then be visualised along with other samples (e.g. in BrainRender.)

{% hint style="info" %} For a tutorial on using this tool to trace silicon probe tracks, see here {% endhint %}

{% hint style="warning" %} For now, this tool depends on amap. Please run pip install amap and then amap_download before running this tool if you don't already have cellfinder installed {% endhint %}

Prerequisites

Data must be registered to a standard atlas (currently only the Allen Reference Atlas is supported) using amap (or the amap based registration in cellfinder). Please follow the instructions for these packages, and ensure that the channel that you want to segment is downsampled (e.g. using the --downsample flag in amap).

Usage

Command line

    manual_seg

Run manual_seg and a GUI will open:

Click Load project and choose the downsampled image (in the registration output directory) you would like to segment.

manual_seg will transform your image into standard space (this may take a few minutes) and then display the image in a napari viewer:

{% hint style="warning" %} If you rename anything in the viewer (a track, region etc.), ensure that there are no spaces in the name. Use an underscore (_) instead. {% endhint %}

Navigating

In the window that opens, by default, only one image will be opened (your image in standard space).

You can navigate around the volume:

• Use the scroll bar at the bottom (or left/right keys) to navigate through the image stack
• Use the mouse scrollwheel to zoom in or out
• Drag with the mouse the pan the view

You can adjust the view of your image, by selecting its "layer" in the sidebar (click on Image in standard space), there you can change the gamma enhancement, contrast limits (right click for finer control) and the colormap used.

The buttons directly below the layers can be used to rotate the data, reset the view and view in 3D etc (this may be slow as the entire volume will need to be loaded into memory).

To trace a linear track

• Click the Add track button
• If required, rename the track (by selecting the track_0 text)
• Navigate to where you want to draw your region of interest.
• Make sure the add points mode is activated (by selecting the + symbol)
• Trace your track by adding points along it. You can add as many, or as few as you like, and this can be done in 3D by changing the viewer plane as you go along.
• Repeat the above for each track you wish to trace
• Save the tracks Save making sure that the terminal window shows that the saving has completed.
• The points can then be joined using spline interpolation by clicking Trace tracks. You can change:
  • Fit degree- what order spline fit to use (the default is 3, cubic)
  • Spline smoothing - how closely or not to fit the points (lower numbers fit more closely, for a less smooth interpolation)
  • Spline points - this doesn't affect the interpolation, but determines how many points are sampled from the interpolation (used for the summary)
  • Summarise- defaults to on, this will save a csv file, showing the brain area for each part of the interpolated track (determined by Spline points )
  • Add surface point If the brain surface is damaged, you may not be able to trace perfectly from the surface. Selecting this option will add an additional point at the closest part of the brain surface to the first point, so that the track starts there.
• Once the tracks are extracted, the fits will be overlaid on the image, but you can preview them in 3D in brainrender (see below).

To segment regions

• Click the Add region buttons

• If required, rename this region (by selecting the e.g. region_0 text)

• Navigate to where you want to draw your region of interest.

• Choose a brush size (top left box)

• Make sure painting mode is activated (by selecting the paintbrush, top left). You can

  go back to the navigation mode by selecting the magnifying glass.

• Colour in your region that you want to segment, ensuring that you make a

  solid object.

• Selecting the ndim toggle in the top left will extend the brush size in three dimensions (so it will colour in multiple layers).

• Add a new region if required (Add region)

• Repeat above for each region you wish to segment.

• Save the regions Save making sure that the terminal window shows that the saving has completed.

• Click Analyse regions to analyse the spatial distribution of the regions you have drawn.

  • If Calculate volumes is selected, the volume of each brain area included in the segmented region will be calculated and saved.
  • If Summarise volumes is selected, then each region will be summarised (centers, volumes etc)

Editing regions:

If you have already run manual_seg, and run it again, the segmented regions will be shown. You can edit them, and click Save to resave them.

Previewing in brainrender

Segmented regions (once saved) can be previewed in brainrender in 3D. Simply click on View in brainrender. There are a number of options you can change:

• Segmented region alpha Transparency of the regions you have manually drawn
• Atlas region alpha Transparency of the brain regions rendered
• Region to render Optional atlas brain regions to render in addition to the manually drawn regions
• Segmented region shading Shading type for the regions you have segmented, choose from flat, giroud or phong.