Trying to trace neurites within a limited radius of an object

CNTNAP2 · October 5, 2016, 4:08pm

Hi,

I’m trying to design a pipeline in which I can:

Identify the center of a nucleus
Superimpose a circle with a fixed radius (say, 50 microns) on the image, using the center of the nucleus as the center of the circle

I am hoping to do this in order to trace neurites out a given distance from the center of the cell.

Does anyone know if this is possible in CellProfiler, or is it better to employ ImageJ for steps like this?

Many thanks,

RFM

CNTNAP2 · October 5, 2016, 4:29pm

Alternatively, if there were a way for me to manually measure the distance, say, from the soma to a given distance away from the soma (e.g. 25 microns), that would also be helpful.

David_Logan · October 5, 2016, 4:30pm

Briefly:

ID Primary on the nuclear channel
ID Secondary on the neurite channel. The “Distance - B” method has a setting called “Number of pixels to which to expand the primary objects by”. Set this to the number of pixels equivalent to 50 microns or whatever.

HTH!

CNTNAP2 · October 5, 2016, 5:02pm

Hi David,

Thanks so much for this; I am glad to know about that option!

Unfortunately I think my situation is somewhat complicated. I’m using projections of confocal stacks, and the images are such that identifying secondary objects has only worked in the Watershed mode. When I’ve tried other “identify secondary objects” methods, they just haven’t worked unfortunately.

I will upload my pipeline and links to a sample image set here

Overall my goal is to be able to measure intensity levels of GFP and RFP in neurites within a given distance from the soma. Right now the pipeline does not include that capability, but I am trying to figure out how to add it. I am totally fine with doing manual editing as a part of this, if needed!

Many thanks,

Ruth
mGluR_Confocal_9-23_Version3.cpproj (1.0 MB)
https://drive.google.com/open?id=0ByycYvo_VTB4eUFyY1dYeXhmWHM
https://drive.google.com/open?id=0ByycYvo_VTB4M2ZPaXVHdzMyYjg
https://drive.google.com/open?id=0ByycYvo_VTB4WUdIaE1HVER1MVU
https://drive.google.com/open?id=0ByycYvo_VTB4eFl3VlFNRjZCMkk
https://drive.google.com/open?id=0ByycYvo_VTB4c0kwZFJwUzQ4VWs

CNTNAP2 · October 5, 2016, 9:03pm

My colleague actually figured out a nice workaround, with the following sequence:

–> Identify a given defined area around the nucleus (e.g. a 200 pixel area in all directions) by using Identify Secondary Objects - Distance N
–> Save the outline of the resulting object
–> Superimpose the outline onto the main image
–> Identify cells via Identify Secondary Objects with watershed method
–> Manually edit objects to ensure all neurites within the outline are traced!

David_Logan · October 11, 2016, 3:20pm

This is clever! But I really would like to see a cleaner implementation someday. Sholl analysis ought to be implemented someday (measuring neurite crossings at various distances from the soma) and could be adapted to measure imntensity as well I would think.

Side note: I should have pointed out earlier for neurite finding using IDSecondary CellProfiler you really should UNcheck the “Fill holes” option. And also note that the MCT thresholding method was developed for neurites and is usually a good starting method.