Cell distribution in my region of interest

I would like to estimate number of specific interneurons in striatum and I am beginner in stereology. But cells in which I am interested are not nicely distributed.
I wonder how reliable is counting cells in regions, where they are not located in an isotropic manner.
I want to count cells in striatum, but there is a medio-lateral gradient of cell density. I tried to adjust the sampling grid, but when I have a smaller grid I have between 4-6 cells per frame in medial parts of the structure, however in the lateral parts I end up with many empty frames. With a bigger counting frame I have around 20 cells per frame medially, laterally then I have between 1-5 cells per frame.
Is it actually possible to divide structure according to its density and sample medial and lateral parts separately, then estimate the total number of cells by sum of the results? Or should I proceed as I’ve described in the second paragraph?
I wonder how this can affect error numbers. Which numbers should I keep in mind and what value my coefficient of errors should have?
Thank you.

This is a very common issue that people run across.

Stereology will work to count cells even if they are heterogeneously distributed. As you’ve seen, you will need to spend some time figuring out sampling parameters that will work for your tissue.

I would recommend setting your parameters to be appropriate to count the sparsest area, then increase your grid size or decrease your counting frame size to count fewer cells.

So, your process looks like this:

  1. Put up an average sized section
  2. In the Optical Fractionator workflow, trace your ROI (only this one section for now).
  3. Go up to high magnification (oil) and move to a somewhat sparse area. Pretty much anywhere will work for this, and we are going to change these parameters later so don’t worry too much about it for now.
  4. Set your counting frame size to hold 3-5 cells.
  5. Set your grid size: Use the “enter number of sites” option, and enter 10 sites.
  6. Count this whole section with these parameters. After you’ve finished, consider:
  • How many markers did you place in total? Note that you can get this from the results, but you can also right click the marker toolbar and turn on show marker summary to get this information
  • How many markers did you place on the sites with the most cells marked?
  1. Change your parameters as needed:
  • If you marked more than ~20-30 cells and you plan to count 10 sections per subject, increase your grid size to mark fewer cells next time. If you marked fewer cells than ~20-30, decrease your grid size to mark more next time.
  • If your sites with a ton of markers had more than ~5-6 markers placed, decrease your counting frame size.
  • Changing either grid size or counting frame size will impact the number of markers you placed, so you will likely have to tweak both of these numbers. For example, if you decrease your counting frame size so that you sample fewer cells per site, you may also need to decrease your grid size to sample more sites.
  1. Now, proceed to count an entire subject with these parameters. You should be aiming to sample around 150-200 cells with a CE (Gundersen m=1) of 0.15 or below; continue tweaking your parameters if needed. Note that it’s fine to change parameters between subjects. But, you cannot change parameters part way through counting a subject – if you start counting and realize these parameters won’t work, you’ll have to change them and start over from the beginning of that subject.

A very common misconception is that you can’t have counting frames with no cells. That’s not true! It is completely fine to have counting frames with no cells marked, and for some types of tissue like yours, or BrdU labeled cells, many or even most of your sites will have zero cells marked.