Hello

So I have mutant mice that misproject axons in the spinal cord. Essentially, axons enter a territory that they are not supposed to be in.

I want to image the spinal cord as a Z-stack. I get images like this for the wild type (image B). In my mutants that nice clean line of neurons misprojects where it is not supposed to.

https://goo.gl/images/8lY24a

So my Z-stack is essentially 100 of those images.

I want to automate scoring each image, through the spinal cord, so I can do statistics.

This procedure is complicated by the fact that the spinal cord is not the same along it's length. It varies in shape (so if you defined a 1cm2 'neurons normally absent' area 5mm from the bottom of the section, at one end, it might not be exactly the same at the other.)

Additionally the misprojection phenotype might be stronger along one part of the axis, so I want to be able to compare how different chunks of spinal cord vary along the axis, within the embryo and between embryos.

This figure:

http://www.nature.com/ncomms/2015/150511/ncomms8028/fig_tab/ncomms8028_F3.html

Here is the sort of idea I was thinking of (grid pattern), scoring. But this paper has the issue that it is only picking 2-6 samples per embryo. As there is fluctuation in phenotype along the spinal cord axis (and probably between embryo ages), and the experimenter is placing the grid, I feel that this is very susceptible to bias.

If anyone has any ideas for how to approach this problem, or knows of programs/scripts/FIJI add-ons that might be useful, please let me know.

Thank you!

Christine