Problem with save images of outlines -images dark

Hi, I have a workflow (attached) that is analysing mitochondria in images from a PE operetta.
I believe they are 12 bit in a 16 bit file, so I apply a rescale intensity first with the recommended settings for 12bit.
The problem I have is with the overlay outlines step. I try to do this on a falsecolor image generated from the two input channels after the rescale intensity step. The falsecolor image seems fine in the graytocolor module (mito in red & nuclei in blue):


but when I use it as a color background for the overlays, and save in png or tiff etc, I only see the red channel, and its dim:

What am I missing ??

Thanks again for this great tool & forum!

YR_070909_binary-4.cp (26.4 KB)

Hi just a note on the workflow, I modified it to my purposes from a previous post on the forum - if it looks familiar!

Hi Keith,

What happens to the outlines if you skip the rescaling step and make sure “Rescale intensties” is checked in LoadImages? Just curious…

Hi, rescale intensities is set. I will try without it. Any other ideas?

To clarify: Remove the RescaleIntensities module but make sure the “Rescale intensities” in LoadImages is checked.

Another question regarding this workflow, I would like to improve the segmentation of the cells (cytoplasm), identify secondary step.
I get strange results with the propogation method, where there are strange edge effects of the operation. I have tried several combinations of the threshold correction & the regulatization factor, and have tried increased amounts of gaussian blurring & morphological opening operations before the intentify secondary step, but I haven’t really found a good solution:


Perhaps an addition to the propogation parameters to add some morphological open/close steps at the end would help remove the strange edge effects. I could do this in separate steps, but if there is a better solution let me know!


Just a follow up on the initial problem.

For the overlay image I still have the same problem, I see the blue nuclei in the false color image only if I turn on the log normalised view.

[attachment=1]log normalised to see blue nuclei.png[/attachment]

with just the normal mode I see the following:

[attachment=1]log normalised to see blue nuclei.png[/attachment]

the script is the same as the one above, without the rescale intensity steps…

Hi Keith,

OK, this bears further examination. Could you upload some sample raw image data, plus your pipeline thus far?

Hi Matt, I attach 2 input images and the workflow.
I think I more or less solved the issue I had with the identify secondary having jagged strange edges, I use now the distance B option, which seems to work well with these images.

thanks! Keith

YR_070909_binary-10.cp (31.6 KB)