Identify

cellprofiler

#1

Hello All,

I’m one of the new and excited users of CellProfiler.

But since a couple of days I’m struggling with the IdentifyPrimAutomatic-module and especially with the threshold correction factor and bounds.

Are there any handful hints and tricks to initialize these values? And what about the others?
Just playing around, many trial-and-errors did not lead to acceptable and stable results.

An example of one of the images that we have to quantify, mostly with the MeasureObjectAreaShape-module, can be found on
[]()

Any idea or method will be greatly appreciated. Thanks in advance for sharing!

Greets,

Wies


#2

Hi Wies,

CellProfiler was initially designed to work with IF images. We are currently working on getting some modules to identify bright field images. Can you get IF images of your cells? This is the reason it is probably so difficult for you to find a good threshold correction and bounds for your images. Thresholds work by separating your objects (cells) from the background. In IF, your cells are always much brighter than your background. In bright field images, the background is generally the same brightness as your objects.

I will make an announcement on this forum as soon as we have some working modules for bright field images. Sorry I could not be more help!

Mike


#3

Hello Mike,

It’s indeed a pretty difficult task to quantify BF images.
For this project fluorescence is not suitable, so one has to be inventive…

For those kind of images I used Sobel edge detection before running them through CellProfiler, with success :smile:
(I just needed to measure the cell area)

Thanks again for this nice piece of software, and I’m looking forward to those new modules.

Greetings,

Wies


#4

Thats great! The next release of CellProfiler will contain a “FindEdges” module which will have all of the edge finding algorithms (Sobel, Roberts, etc.) availble to use in CellProfiler, which should make it less tedious to do before hand. That should be out soon.

Good luck,
Mike