Optical Fractionator making me oversample?

I’m using the optical fractionator and have a few questions:

  1. I’m using the same sampling parameters for all of my samples. I want to sample 5% of my region of interest, but the program seems to be making me count more than 5% for about half of my samples (the total number of sampling sites is higher for those animals despite having the same or lower number of sections).

  2. I’m counting cells from about 15 animals. Do we need to have the same number of sections for each animal?

  1. It is not the number of disectors that is held constant but rather the size of the disector and the size of the space among the disectors (grid step). You are using 5% or 1/20th as your Area Section Fraction (ASF) (counting 50um x 50um). Here is how that might look for a section of a given area:


    Given the size of the cross section and the random throw, we get 61 sites.
    Here is how this looks for a section that is about ¼ the size in cross sectional area:

    The section is smaller so there are less disector sites (21 sites).
    This seems like a lot of sampling. We recommend starting with a pilot study set up such that you are using ten sections, a mean of ten disectors per section, and a disector that will ‘catch’ 0 to 5 cell tops. Take a look at this webinar for more information on pilot studies: https://www.youtube.com/watch?v=QNH4vDCou4E&list=PLpEyqmHMIjwm75lawT1lMXNzJLBBGD758&index=4
    You may have over-sampled. To find out take a look at your CEs. They should be below 0.1, but you may be doing too much sampling if they are considerably below 0.1. You can use the over-sample/re-sample feature in Stereo Investigator to examine the population estimate when the disector or section factor is reduced. If the results show that you get the same estimate with half the number of disectors, then you can consider using half the number of directors to improve the study efficiency.

  2. No, there is no requirement to have the same number of sections per animal nor the same number of disectors per section within an animal. In fact, it would be quite a coincidence if all your animals had the same size anatomical region of interest. And it would be a very regular region that would give all sections of the same area. There will be biological variability causing different numbers of sections per region from different animals. Within a single animal, the region size on a given section will be different throughout the volume of your ROI causing a different number of disectors to be used (see question 1) on each section.
    Scientists have a drive to normalize. However, with stereology, our goal is to keep the interval of sections counted constant in addition to the ratio of the disector size to the grid-step size. If we force the number of sections or number of disectors to remain constant, we might not sample the whole region in some cases, introducing the potential for biased results.

1 Like

Thank you Aidan, I just want to add a NOTE that if an animal was oversampled, it can still be used in the study; but you can consider sampling less on the subsequent animals in that group.

1 Like