Segment neuronal growth cone and neuron tip

Hi Everyone,

I’m trying to segment neuronal growth cones as will as the tip of the processes that holds the growth cone. The idea is that the size and shape of the growth cone dictates it’s health but this metric needs to be compensated by the size of the tip holding it. So far I have a decent pipeline that cleans up the images and segments the right parts. I plan to use the output of the skeleton/endpoints morph module as input objects then segment/measure those object areas on the right channel to get the growth cones. The issue I have is how to find the width of the processes tip holding the growth cone… Put another way, if you were holding a cup in an outstretched hand, I need to measure the size of the cup and the diameter of your wrist. The ratio of the two would tell me if your wrist could hold a bigger cup, but isn’t (weak growth cone) or if your wrist is holding a decent weight cup for it’s size (healthy cone).

Attached is a sample image set plus pipeline.
Images :

  • 380: Dapi staining neuronal nucleus
  • 555: Phalloidin staining everything
  • 488: BetaIII staining neuronal processes

Use DAPI as input primary objects, use BetaIII with skeletonization and endpoint morph commands to find endpoints, use Phalloidin as channel to measure growth cones. (558 KB)


There is no straightforward way (yet) to measure neuron width. However our colleague Carolina Wahlby says that she uses this method on C. Elegans worm:

  • Use Morph to create a distance transform of a binary (representing all objects)
  • Use MeasureObjectIntensity with the distance transform image as input. The Max intensity and upper quartile give good approximations of width

See the attached pipeline, in which I added a few modules to do the above. It would benefit by you defining a restrictive region of the axon within which to measure the width (“Where along the neurite do you want to measure the width?”). The output is only an approximation, I’m afraid, and we are planning to add the width (at least average width) along the neurite, but we have not started this yet.

Here are some other things I noted while inspecting your pipeline:

  • Rescale Intensity (both) - likely not needed
  • IdentifyPrimaryObjects - try unchecking “Discard objects outside diameter range”. Again, I would consider using the “soma” image as input here, and not the Rescaled_soma, since it appears saturated
  • IdentifySecondaryObjects - try unchecking “Fill holes…”. This often finds more veridical neurite borders without filling gaps spuriously.
  • The two Morph modules look to be duplicates (OK, it just wastes resources)

I hope this helps!
growthcones_DLogan.cp (18.4 KB)