Flex files in a z stack

Hi,

I’ve just started playing around with CellProfiler as it looks like a great piece of software,
I would like to explore it as an alternative to Acapella which came with our Opera microscope, however I am having some trouble understanding how to deal with my images.
What I normally do is load a flex file that will contain GFP and RFP channels and 5 stacks in each, I have managed to load the flex files using the latest trunk build of cellprofiler but how do i deal with z-stacks? it seems to load as a single image.
I only want to identify objects in the mid-section of the cells, but i would normally measure variance in pixel intensity across all images and choose the image with the lowest variance in intensity as a way to find the midsection in the stack as it can vary well to well in a high-thoughput screen. This image would then be used for cell image analysis.
An example of the image files i use can be found here: http://www.qfpost.com/download.do?get=9e22847379537cfe1dcc5dd4001f1cb2

any help would be much appreciated

Kind regards
Peter

Hi Peter,

For your application, you would need to use "tif,tiff,flex,zvi movies’ under file type in LoadImages. At that point, you should be able to add channels for GFP and RFP. Each pair of image channels for each value of Z will be loaded and processed for each cycle. You can then add a MeasureImageIntesnity module to calculate intensity values such as standard deviation. You can use this as a surrogate for the variance, or use the CalculateMath module to obtain the variance as the square root (i.e., ‘None’ as the operation, and raise the power to 2) of the standard deviation. An ExportToSpreadhsheet module will let you export the values.

The problem comes in using this variance value to inform your decision in which Z-slice to use. You would need to calculate the variance in one pipeline first and then use that value in another pipeline for identification. Unfortunately, there is no way to select a Z-slice automatically outside of test mode. However, what you can do is use the FlagImage module to skip an image if a measurement doesn’t meet a criteria you specify. So if you know variance/STD cut-off value, you can then use an identification module that is run only if the criterion is met. I’m attaching a pipeline which sketches out what I’m describing.

Hope this helps!
-Mark
2011_06_24.cp (7.84 KB)