Hi Ana, 

I'm not sure how to quantify the degree to which the projections are neat and parallel, but if there are physically more projections in one treatment than in the other treatment, might you be able to quantify the difference in "density"  between the two conditions?  

I'm assuming that these are axon projections? Here is a link to a J Neuro Methods article that describes how overall density of projections can be measured in FIJI: https://www.ncbi.nlm.nih.gov/pubmed/16469388 .  I've used a modification of this recently to quantify differences between groups in 40micron sections with fluorescently labeled axons, obtaining z stacks with a spinning disk confocal similar to the situation that I think you describe (you'll have to modify the parameters in the article to fit your acquisition settings). The basic idea is to acquire your z stack, create a max projection, and calculate the amount of area of your image that is taken up by the projections that you're interested in. You should randomly sample several times within each tissue section (across the entire brain region), then do the same thing on several more tissue sections of the same animal (i.e. stereology), and create an average density count for each animal across all of the images you obtained. Repeat across all of your animals/conditions (keeping all acquisition settings AND parameters within FIJI the same), and now compare your groups statistically. 

The problem with the above technique is that it is not a very accurate portrayal of the axons in 3 dimensional space - you are flattening a z stack. You describe all of your axons being "in parallel" in one condition, while in the other there are more projections (axon collaterals?), so perhaps the above technique will do the trick. A second more sophisticated option though is to use true stereology to quantify the actual length of all projections randomly sampled across your section, and this can be done in 3-dimensions. I am not aware of any FIJI plugin that can handle this, but take a look at the "Spaceballs" probe that is included in software such as Stereoinvestigator (from Microbrightfield, the same company that makes Neurolucida).  Here is a youtube video of the technique https://www.youtube.com/watch?v=nvMM1CRR7K0 .  There might even be a way to assess the parallel vs non parallel features you describe.  

I hope that spurs some ideas!

-Dan 

Thanks a lot for your time and help, Dan! I'm going to check that. I thought about something like that, but as an histogram of pixels used for the projections. I'm not sure what kind of projections are (axon/dendrites, or more cells or more projections of the same number of cells, etc). It was not my first question, but I noticed this difference between two conditions....and now I want to know what is happening

Good luck with your stuff :)

Ana