Sholl analysis circle help

sholl-analysis
Tags: #<Tag:0x00007fd540668240>

#1

How do you make the concentric shells on Sholl analysis visible? Also, how do you make the shell the same width?

Thank you in advance for the help!


#2

Right now there is really no option for it. The plugin would have to be modified to allow it. Actually, if I recall properly, a primordial version of the plugin allowed it, but that option was intentionally removed to stimulate the usage of continuos (1-voxel increment) sampling (there is really no good rationale for not using it).
Put perhaps the Sholl mask option already tries to accomplish what you have in mind? Otherwise, if you really feel strongly about it, you could use the Concentric circles plugin by @Wayne.

Sorry, I really do not understand the question. Can you elaborate it a bit more?


#3

The concentric circle pluggin worked. Thank you. But does the concentric circle pluggin generate the same information that the sholl analysis analyzes (meaning will the data be the same?). Also, I am trying to merge the red and blue channel in order to perform sholl analysis. But the program wont let me because sholl can not be performed on a RGB image. I am trying to perform sholl on a neuron and plan to use DAPI (blue) to define the center and use the red channel to set the most distal point. Please help.


#4

Dear waltz, I really do not understand the question. The concentric circle plugin does not generate any data, it just overlays a grid of concentric circles over an image. Which data are you referring to? Manual counts? Importantly, you still haven’t explained why you would like Sholl shells to be rendered. Please detail the problem you are trying to solve, and we’ll all try to help.

Can you provide a representative image? It is extremely difficult to comment without it.

As for the merging of channels, have a look at the documentation. Also note that you can always use the command Command Finder/Launcher to find out what commands to use. E.g., type L, to activate the Command Finder/Launcher, then type merge or channels to find out which commands can be used to create multi-channel images from individual grayscale images.


#5

You haven’t followed up, so I’m assuming the issue is solved?
Also, I just wanted to mention that the new version released today (v3.6.4) supports multichannel (composite) images, so that you no longer need to split channels prior to analysis. Note that lack of RGB support remains intentional, as explained in the FAQs.


#6

Im sorry I am verge with my questions. Im still trying to understand the pluggin. But I am trying to analyze primary neurons using sholl analysis to see if the treatment that has been added to the dendrites affects their complexity. When I do sholl analysis I would like to see the circles that are analyzing the number of intersections. Is there a way to show the circles instead of showing the colorful neuron that is shown in the sholl mask?

Also I would like to measure consistently so is there a way to change the measurements from pixels to microns? All is there a way to know the the distance each distal neuron is from the center?

For my comment on the red and blue channels, I merge those two channels in order to use the DAPI channel as the center of the analysis and the distal part of the dendrite as the end radius. I was wondering if there was a way for me to analyze the merged RGB image with sholl analysis?


#7

In cases like this it’s best to post a representative image. It makes it easier to reply with concrete answers, and allows readers to immediately relate to the problem without having to go through long posts (like this one) just to understand the issue at hand. For clarity, I’ll split your post into a list of questions:

How to display sampling shells as circular ROIs?
Last saturday I released v3.6.5 that adds a new Overlay 2D shells checkbox in the Output section of the main prompt to allow this. Please run Help>Update… to fetch the update. A couple of things:

  • ROIs are added to the image overlay as either circular ROIs, or composite ROIs when using hemicircles, and are named according to their radius.
  • Thickness is defined by the value specified in # Samples. Their color by whatever selection color you have specified in IJ, i.e., the one reported in the (Multi)Point Tool, in the main IJ window.
  • Right now it is limited to 2D images (which I’m guessing you are using). It is not immediate to me how to extend the rendering of these ROIs to 3D images.

How to perform the analysis in pixels in spatially calibrated images?
If your images are spatially calibrated, analysis is performed in physical units, otherwise in pixels. If you want to impose distances in pixels: Run Analyze>Set Scale…, click on Click to Remove Scale button, than activate the Global option. The Global option is explained in the user guide. Note that this is an ImageJ feature, and is not specific to the Sholl plugin.

NB1: I would strongly advise against disabling spatial calibration on 3D images. If you do so, sampling distances will be calculated assuming an isotropic voxel size, even if, your images are likely anisotropic due to the lower axial resolution of conventional microscopes.

NB2: You shouldn’t need to discard an image’s metada to achieve consistency in a processing workflow. Perhaps some of the procedure(s) you’re using should be revised?

How to analyze RGB images?
You convert them to multichannel images, and then run Sholl on a selected channel. I was convinced I had made this point clear when you prompted me to release v3.6.4 last month. I really regret you are still having problems with it. In short:

  1. Open your RGB image
  2. Activate the Channels Tool by pressing Shift+Z (Image>Color>Channels Tool…)
  3. Choose Make Composite Image from the More>> dropdown menu
  4. Set the Display mode to “Grayscale”
  5. Activate the channel to be analyzed using the “C” slider underneath the image canvas.

NB: Step 4. is required because “Composite mode images cannot be tresholded” (This is the actual error message you’ll read in the status bar when trying to segment a multichannel image displayed as “Composite”).

However there should be no need to use RGB images. You can either 1) use your DAPI image to define the central point ROI and transfer it to the analysis image using Shift+E (Edit>Selection>Restore Selection), or 2) Merge the single channels directly into a Multi-channel image using the Create composite option in Image>Color>Merge Channels…

Was this helpful? It would be great if you could update the documentation page with this info (Perhaps in the FAQs), so that others can profit from it.

As per, [quote=“waltz, post:6, topic:1564”] is there a way to know the distance each distal neuron is from the center [/quote]

I have to admit I dot not understand the question. Could you elaborate? What is not working for you?


#9

Good Morning @tferr ,

Sorry to bother you but I need your help.

I recently learned that simple neurite tracer provides the option to use sholl analysis, however is it only used for 3D images? Will it work on 2D images of primary neurons?

I use the averages of the sholl profile to create a graph for analyzing the neurons, however the results of the sholl profile seems to skip numbers, which I end up adding 0s for numbers that were skipped. For example, I would set up the measurements as the radius step size is 5.0, the starting radius is 2.0, and the ending radius is 1500, but the problem is that when I review the results numbers are missing: instead of the x values going from 862, 867, 872, 877…, the results show 862 then the next value being 992. I usually add 0s for the y values that were skipped. Is there a way to fix this problem?

I recently discovered that there is a sholl analysis pluggin found on the Ghosh lab website. How does it compare to the sholl analysis that is already installed on image-J?

Thank you in advance for your help.


#10

It will work with either 2D or 3D images. See documentation and detailed tutorial.

Version 3.6.10 released yesterday lists zero values in the detailed table. I never consider this to be a problem and assumed zero-padding to be trivially achieved after running the plugin. The rationale for ignoring zero-values is explained in the FAQs.

This is also answered in the FAQs.


Display "0" values in Sholl Analysis plots