Z Stack Analysis

Hello everyone,

I’m a research assistant in a lab that currently primarily deals with 35 micron thick tissue slices, stained using antibodies (it’s pretty standard immunohistochemistry procedure). One unexpected question we were posed during a recent presentation we did was, “How do you (we) account for differences in tissue thickness when generating max projections from Z-stacks?” We took a look at some of our .lif files (generated using a Leica confocal microscope) and realized that the number of “slices” containing signal differed between regions of interest, and the first slice containing signal differed as well. For example, ROI#1 may contain signal in slices 4 through 24, but ROI#2 had signal in slices 1 through 25. These stacks would represent different volumes of tissue, and as such different numbers of cells, and the whole purpose of my project is to develop an automated image processing/cell counting procedure. The grad student and PI of my lab agreed it would be best to set the slice acquisition range very broadly; i.e., we’d have “blank” slices at the beginning and end of every stack. What I’m trying to figure out is:

  • Using FIJI, what is the best strategy to exclude blank slices from a max projection, while making note of how many slices were included so we know what the volume of the tissue is? Is there an automated option, or will I have to manually decide the slices range for every max projection?

  • If I do have to manually decide what range of slices to use, should I keep that range the same for all ROI in a given sample? Following the example given above, maybe I only use slices 5-23 for ROI#1 and #2, because I know every slice in that range will contain signal.

If possible, I’d appreciate if anyone can provide me with some examples in the literature where Z-slice range decisions were made, so I can learn from other’s example, rather than begging to be spoon-fed on an image analysis forum! Any other useful references would be much appreciated as well; the more I learn about automated image analysis, the more I respect its utility in the modern lab, so the more I want to learn

Thanks so much!

Matt

1 Like

Hi Matt (@JungleJim4322),

I don’t think you’ll need to exclude slices for a max intensity projection (MiP) (but if you want average projection, then what you describe is a concern). Since the algorithm takes the max over all slices, removing blank slices will not influence the result at all

(assuming blank means - “has zero intensity everywhere” and all non-blank slices are positive)

Write back if I’m overlooking something.

  • I see; are you using the max-projections to do the cell counting?
  • Are you computing a MIP for each roi?

:clap: :smile: :clap:

Hope this helps, but please continue the conversation if I’m missing something,
John

Sorry, I left out a key piece of info. You’re correct that, in generating a max projection, blank/negative slices don’t affect the appearance of the final image; however, the number of slices does affect the volume of the tissue. So, in later analysis, if I have a max projection made of 45 slices but 15 are blank, the automated volume analysis will count all 45 slices as containing tissue, giving us an artificially low cell/um^3 density.

Additionally, even if I just want a cell count, I can’t really compare a max projection made of 27 slices to one made up of 33 slices. The 33 slice max projection represents a larger volume of tissue, and therefore more cells, so I couldn’t confidently say, for example, a high fat diet influences microglia number if I don’t even have the same number of cells to begin with.

@JungleJim4322,
you can try to count total intensity of each optical section; as result you should get a bell-shape curve, where X is an index of the section and Y is sum intensity. After that you threshold this curve with a coefficient to find indices of the starting and finishing section for your cells. Clip away the sections below threshold and make max intensity projection. Depending of your data (for example, when staining of cells is on average the same and constant throughout the cell), perhaps the sum-intensity projection may also be suitable.