FIJI and CP different appearance

imageformat

#1

Hi all,

I’ve just started to use CellProfiler and I think it is a great tool for automatic segmentation. My images are z-stacks of 4 different channels, where I would like to study colocalization. Thus I’ve downloaded the CP 3.0.0 version and followed the 3D Monolayer tutorial, as well as Colocalization Example in order to have an idea of how to proceed with this type of files.
I have now started with my images, however I have already found a problem. When I drag my images into the file list either they do not have the same appearance as when I open them by FIJI (see attached screenshot) or they look the same as in FIJI in the beginning but in any case (in both cases actually) they look different from the FIJI files when I start applying the different filters. That is to say, the images I want to start processing always have much higher background than my actual images. They are 12 bits images acquired with a confocal Olympus (.oib files) saved as .tiff with FIJI when I split the channels and gray-scale them . This problem happens even if I transform them to 8 bits images beforehand. Can please anyone help me?

Thanks!


#2

Hi,

One simple explanation could just be the rescaling of the display- can you check your rescaling preferences in the File->Preferences menu and set them to raw? Otherwise, can you upload your images here (or elsewhere like GDrive or Dropbox and link them here) so we can attempt to reproduce?


#3

Hi,
I’ve changed the rescaling preferences (CellProfiler preferences>Intensity normalization>raw) and I obtain exactly the same images as before even if I restart my computer. I’ve tried also increasing Maximum Memory for Java (from 512MB to 6092MB) in case that was the problem, but still there is no better result.
Here you have the link to the photos:

Thank you for helping!


#4

Hi,
So it turns out the reason is that there was a bug in 3D image loading that was causing the images NOT to rescale to 0-1 upon loading- the display expects 0-1 scaling, hence the strange looking display.

We are working to fix it (and in fact are delaying our next release to make sure it includes this fix since we think it’s a critical bug)- in the meantime, you can still use the images in CP3.0.0 just by adding a RescaleIntensity module per channel at the very beginning of your pipeline, set as follows.

We apologize for the inconvenience, and thanks so much for bringing it to our attention!


#5

Hi,
Indeed, with the RescaleIntensity module it worked fine for each channel.
I’m glad this has helped us both.

Thanks!