CellProfiler mask image to FISHQuant

Image Analysis
,
#1

Sample image and/or code

CellProfiler
CellProfiler1239×688 178 KB

Pipeline version mismatch
Pipeline version mismatch822×195 29.3 KB

SQ_seg
SQ_seg505×790 21.7 KB

not a mask

Details

file:///C:/Users/hhrnc/Downloads/FISH_QUANT__Tutorials%20(2).pdf
I have been following smFISH analysis outline using FISHquant in Matlab. I am attempting to perform robust segmentation of cell outlines. I have been following this protocol: file:///C:/Users/hhrnc/Downloads/FISH_QUANT__Tutorials%20(2).pdf
HOWEVER, things delineate in CellProfiler because I had to make my own pipeline which is pictured above to get ‘masks’ to export to Matlab to use as an outline in FISHquant. The reason I had to make my own pipeline is because I get the following error message when I try to use to pre-designed pipeline I get a pipeline version mismatch error message (pictured above).

When I try to use the masks created in CellProfiler in Matlab I get the error message that my tiffs are “not a mask for cells”. Please help. I have tried for so long. I have updated my Java and and am using the most recent version of CellProfiler (4.0).

I want to create automatic cell outlines for smFISH analysis in FISH-Quant.
I am new to CellProfiler and FISH-Quant.

#2

Your PDF attachments of the manual did not come through, sorry.

If you have an idea of what specifically FISHQuant needs (binary images? label matrices? outline images? something else?), and if you upload your pipeline, that will be the best way to figure out where the disconnect is.

#3

smFISHpipeline.cpproj (723.6 KB)
FISH_QUANT__Tutorials (2).pdf (2.7 MB)
CP_pipeline.cppipe (15.2 KB)

Hey Beth! Thanks so much for the help. I have uploaded the FISH-Quant tutorial (hopefully it works this way), my pipeline (smFISHpipeline) and the example pipeline that I could not get to properly load into CellProfiler (CP_pipeline). Thank you for the quick reply!

#4

Also, I believe FISHQuant wants outline images.

#5

Hi @HHrncir,

Looking at the FISH-QUANT documentation, I’m reasonably confident that FISH-QUANT is looking not for outlines of the cells, but rather for a grayscale image where the pixels that define one cell have the same value as each other (and a different value from background, which is 0, and every other cell), as described here:

Screen Shot 2021-02-16 at 1.52.16 PM
Screen Shot 2021-02-16 at 1.52.16 PM765×153 12.3 KB

For that reason, in the ConvertObjectsToImage module, you’ll need to select “Grayscale” instead of “Binary” for the image type. A binary image will make all the objects have the same grayscale value, instead of unique values.

Screen Shot 2021-02-16 at 1.54.34 PM
Screen Shot 2021-02-16 at 1.54.34 PM846×358 75 KB

In addition, in the SaveImages module, you’ll want to select “Image” for “Select the type of image to save.”

Screen Shot 2021-02-16 at 3.18.16 PM
Screen Shot 2021-02-16 at 3.18.16 PM2244×952 211 KB

(As a side note, in your pipeline, you do not need to use the MaskObjects module. This module is intended to remove objects that fall outside of a region, such as if you had a mask for “in focus” parts of your image, you could apply that to your cell objects in order to keep only the “in focus” cells).

Once you’ve made this change, your resulting masks should have individual cells that look different from each other and can be loaded into FISH-QUANT.

Example mask image: ignore the cell shapes, I used one of my own example images that won’t match your data
20210122_CY3_cell_MASK.tiff (1.3 MB)

Let us know how it goes!
Pearl