BrainVoyager Support

documentation overview

  • Increase font size
  • Default font size
  • Decrease font size

Segmentation Helper

When to use the Segmentation Helper

The main motivation for this tool is to provide a way to go back to previous segmentation states when making an error in the manual segmentation. The segmentation history is not automatically stored, but you can easily add a state to the history list with a simple button click.

There are also some experimental functions available to help you to get a good initial segmentation.

The Segmentation Helper is not a replacement for Brainvoyager's Segmentation functions but is meant to work together with the functions available in the Segmentation Tab of the 3D volume tools. It is also recommended that you have both windows open at the same time. You can then for example choose to draw something (using the segmentation tab) and store it in the segmentation helper afterwards.




 Segmentation Helper 1.2

How to use the functions

For all buttons there is a mouse Tool-tip text so just hold the mouse over a button to get some short info!

The Segmentation Helper reads in the VMR when you start it. The only way to be able to return to a completely unsegmented VMR is to start the segmentation helper before anything is segmented!
Also: If you open a new VMR, you have to restart the Segmentation Helper

Segmentation History:
When you are happy with an intermediate segmentation result, you can add it to the end of the list by clicking the add button. You can browse through the list of states by using the radio buttons and the VMR will immediately be updated. It is not possible to remove specific states from the list, but you can always use the "Set Current to State 1" button which will set the currently displayed segmentation as "State 1" and removes all other states.
The Reset Button will also delete the whole history and display the Original VMR as it was at the start of the plugin.

Experimental Tools:
Estimate Brain Segmentation Limits 
This function always automatically works on the original VMR, no matter what the current segmentation state is. It tries out different intensity limits for Region Growing, automatically starting from the white matter. When lowering the minimal region growing value, more of the brain will be selected. The idea is that at some intensity value, the region will grow out of the brain and into the surrounding tissue. This estimate function tries to find the biggest intensity range for which only brain voxels will be marked by the region growing. The range is specified by the From-To intensity values that are automatically shown in the plugin window when the estimation is complete.
The output usually results in a mostly White Matter segmentation but is only meant to give you a good initial estimate for usable region growing values. Since it is only an estimate, you should really adapt the values a bit depending on the region growing output as is described next.

Grow Region from WM
Automatically detects the white matter and performs a region growing from there, using the intensity limits in the "From" and "To" boxes right above the button. The region growing function always automatically works on the original VMR, no matter what the current segmentation state in Brainvoyager or the plugin is. This makes it very easy to try out different region growing values. Simply change the lower and/or upper limits and click the button again without changing anything.
Usually the automatically detected range by the estimate function is not perfect. If you for example want the segmentation to move closer to the gray matter you can decrease the "From"-value. This will most likely result in the region growing out of the brain into the surrounding tissue. The Clean Up function can in many cases correct this afterwards:

Clean Up
This function looks at how well voxels are connected to the main White Matter / Brain block. If a region growing slips past the brain-surrounding tissue barrier, this usually happens because of a very small connection through this barrier. The clean up function tries to remove the voxels that are only a part of the main block through these thin connections. One parameter, the "clean up strength" can be set to control this process.
The clean up strength specifies how well voxels need to be connected to the main block. A value of 4, means that a segmented voxel needs to have a connection to the main block that at the thinnest point has at least 4 segmented neighbors out of a possible 6.  A value of 1 only removes voxels that are completely separate from the main block. Choosing very high values will automatically result in a shrinking effect since it removes the outer layer of the segmentation as well.

Expands the current segmentation area one small step using both the "From-To" values and the bounding box. The function only works on the segmentation within the bounding box and also never expands outside the bounding box.

Everything works quite quickly and of course you can always return to your original VMR. Nothing should hold you back from playing around with the different experimental functions. You can also use them in combination with Brainvoyager's Segmentation Tools like the Expand button!

Usage Examples

1. Get a proper initial segmentation 
Use the Estimate button to get started with region growing but always play around with the "From-To" values afterwards. You don't need to return to the original VMR before doing region growing, it always automatically works on the original VMR, even if there is a current segmentation.
- If the region growing goes into the surrounding tissue, you can try the clean-up function.
- If the region growing is a bit too much into the gray matter you can also use the clean-up function with higher strength and maybe expand a bit afterwards when needed.

2. Redo segmentation in difficult areas 
You can delete difficult parts that you want to re-evaluate using the "Clear inside bounding box" function. Then you can increase the bounding box size one click so it contains a bit of segmentation again. This allows you to slowly grow back the problematic area using the expand button a couple of times in combination with "From-To" values that better suit your difficult area.

3. Erase segmentation parts with the drawing tool 
If you go to the drawing mode (3D Volume Tools -> Segmentation Tab -> draw with mouse) you can draw with CTRL - left mouse button. It is also possible to delete parts by using Shift - Left mouse button. When you delete parts they turn black and you don't see the underlying anatomy anymore. There is an easy way to fill the black spot with the underlying anatomy again:
load the segmentation helper before doing any segmentation so it can store the unsegmented VMR as original state -> Do some segmentation and delete a part using the drawing tool -> Store the current state ("Add state" or "Set Current to State 1") in the segmentation helper to store the segmentation -> Now click on the state you just stored (although it is already selected), this will overlay the segmentation on the original VMR which in effect removes the black spot that was created by deleting a bit of the segmentation.

 4. Do a manual Brain Peeling
Click the Estimate button and when the From and To values have been calculated also the Grow Region button. If the region has grown outside the brain then put press the Clean Up button, with a higher clean strength if needed. If there is a bright area inside the brain then it is best to first change the region growing "To-value" to 225 and press the expand button a couple of times. To select the missing Gray Matter you can lower the "From-value" to about 60 (depending on your dataset) and click the expand button a couple of times. When the whole brain has been selected you can click the "Reload Marked" button in the Segmentation tab of the 3D volume Tools to complete the brain peeling.

Version History:

(23-07-2010) 1.0
is the first version of the segmentation helper.

(05-08-2010) 1.1
- added Bounding Box with visualization option
- new expand function
- fixed bug that didn't clear memory after closing

(27-10-2011) 1.2
- reduced memory need for storage of segmentation states with factor 8 (bit instead of byte)
- added support for .5 mm VMRs!
- improved and fixed some Bounding Box functions


You need to have at least brainvoyager 2.1.2!

Here are the download links for the segmentation helper.

Segmentation Helper 1.2 Windows 32 bit
Segmentation Helper 1.2 Windows 64 bit
Segmentation Helper 1.2 Mac OS X (compiled in Snow Leopard)

Please write to support AT brainvoyager DOT com in case you find any errors or want to give feedback on the segmentation helper!