Cell area determination


#1

Hi,
I’m new to Cell Profiler & have been unsuccessful in determining the area covered by cell staining (see attached photo (attached as a jpeg to save space)). A problem is that these osteoclasts do not stain uniformly, but I need to calculate the total area of the multinuclear cells, but do not want to include the area of the mononuclear (stained) cells. I have been unable to set up the correct pipeline and have received a variety of error messages. Any help would be appreciated. Thanks



#2

Hi Wayne,

Thanks for giving CellProfiler a try for your imaging assay. Can you upload your tentative pipeline as well so we can take a look as why you’re getting error messages? Also, if you can cut-and-paste your error messages, that would be helpful too. Then we can see how close you are to getting the pipeline in good shape.

Regards,
-Mark


#3

Hello again,

In order to solve your problem, I recommend a pipeline similar to the following:

  • LoadImages

  • ColorToGray: Split the image into (R,G,B) components. Based on the image you provided, it seems that the green component gives the highest contrast for the next step

  • IdentifyPrimAutoatic: Identify the objects in the image (which I’ll refer to as Cells)

  • MeasureObjectNeighbors: This gives a measure of the number of neighbors each individual Cell has. The idea here is that we can identify the cellular groupings by the fact that the Cells in those clumps will have a lot of neighbors whereas the isolated Cells (presumably mononucleated) will not. Also, give a name to the resultant image of the number of neighbors

Ordinarily, I’d recommend using FilterByObjectMeasurement to capture only those cells which share a suitably high proportion of their perimeter with their neighbors. However, FilerByObjectMeasurement is broken for Neighbor measurements in the current version of CellProfiler, which means another means is needed in order to filter the Cells for multinucleated ones.

A possible workaround is the following:

  • MeasureObjectIntensity: Use the Cells as the objects and the Neighbor image as the input image. This will assign to each Cell the measurement of the number of neighbors as if it were an intensity measurement.

  • FIlterByObjectMeasurement: Use Cells as the input object, Intensity as the category, and Mean or Min/Max as the measurement (which one doesn’t matter so much since the same single value is assigned to each Cell). Then filter by the number of neighbors you feel suitably defines the multinucleated Cells (probably greater than 1 or 2).

Regards,
-Mark