Hello Danielle,

I'm not a microglia expert, but in terms of image processing, I can suggest either one of the automatic Surface Segmentation algorithms of Imaris, or the free software Ilastik, which has nice supervised learning capabilities (you input examples of your structures, and the software attempts automatic segmentation based on that).

Good luck,

Guy

Hi Danielle,

Skeletonizing alone is insufficient as you might see.

You could add other filters before doing that or rather use thresholded thinning instead of skeletonizing.

For instance a median filter before will level out intensity variance and then thinning and skeletonizing will work better.

If you do a thinning you can preserve the somata and get the real length of the processes (and the branching). However, I use Image Pro Premier for this (and don't have experience with Neurolucida). Next I use a limited adapted threshold for defining the threshold of the thinning algorithm (that in the end makes 1 pixel thick objects of all processes so that they can be counted).

And for classification to sub-types I do like Guy before suggested. I train the software using a set of different parameters (textural and common like standard deviation of signal) and then the system runs a factor analysis defining a recipe for cell type separation.

Hope this helps and is not too gobbledygook...

Best

Dan