Some questions about ZVI Images

I have created a pipeline which:

  1. opens a ZVI image from the Axiovision SE64 software
  2. splits said image into four channels (green/BrdU = 1, red/Sox2 = 2, blue/EdU = 3, white/DAPI = 4)
  3. uses Align function to align the split images
    4)allows the user to rotate each channel’s image then crop it with the mouse to determine a region of interest (in my case, the ventricular zone)
  4. saves the cropped images (SaveImages module)

I am running into a couple of issues with my images:

  1. The ZVI file is not the raw image. In Axiovision, I used their software to subtract the background from the images to be analyzed. I then saved these images as a new ZVI. When I load the image into CP, however, the image loaded is the original (with no background subtracted). Does this have anything to do with how CP is loading these images? If not, do you have any suggestions as to how I could load a 4 channel image which has been processed? I tried to export the image as a 4 channel tif from the Axiovision software, but CP gave an error message when I tried to make a pipeline to open this type of image.

  2. I am using the SaveImages module to save these cropped images for later analysis. The images produced at the end of the pipeline, however, are extremely dark. I can hardly see the cells in the images. I am pretty sure that this is because I have no idea what type of file I should save it as and because I’m not sure what type of color map to use. :smiley: I tried saving as tif in 8 and 16 bit and it produced the same undesirable result.

Lastly, I noticed as I was running the pipeline that when the Crop and Rotate windows are open, there is no indicator as to which file is being processed/which cycle is currently being run (as opposed to the Align window or the window that pops up after I have cropped an image). Is there are way for this to be displayed in the Crop/Rotate windows as well? The reason is that CP does not seem to process the images in a particular order (Cycle 1 then 4 then 2) and I need to keep track of which image I am on so I can rotate it appropriately.

If not, is there a way that I can order my pipeline so that I do all of the cropping/rotating for one file before CP opens the next one? (Example, rotate all four channels and crop File A then have CP start with File B).

I have attached my pipeline, an example of the output image that is too dark/wrong, and two ZVI images. Also, sorry for the really long post! :smile:


Load_ZVI_Rotate_and_Crop_Pipeline.cpproj (79.6 KB)
ZVI2.zip (7.29 MB)
ZVI1.zip (7.21 MB)

I don’t know which images you posted, the raw or the subtracted ones, so I can’t tell from that. I also noted that you used LoadImages rather than the Input modules to load the ZVI. I’m posting a modified version of your pipeline which loads them in using these modules. Drag/drop the subtracted ZVIs into the Images module, press the “Update metadata” button in the Metadata module, and it should work fine.

This may depend on the image viewing software that you are using to view it with. Some (like IrfanView, which I use) automatically scale the intensity so it’s visible regardless of bit-depth. Others, like Windows Photo Viewer, don’t do this, so it may appear dark since the pixel intensities may be in the lower range.

That is a good point; I’ve added this to our bug tracking system as a feature to add for the future: github.com/CellProfiler/CellPro … ssues/1167

Yes, you can use just one analysis worker for the pipeline so that the job is not distributed to multiple processors. That way, the analysis proceeds strictly sequentially. You can do this by going to File > Preferences… and setting “Maximum number of workers” to 1.

Regards,
-Mark
pipeline.cppipe (14.8 KB)

Thank you for the prompt response!

I had posted two different subtracted images originally.
I used the pipeline that you gave me and am still running into the same issue. I have attached two versions of the same ZVI file - one raw and one subtracted. I ran both of these images through my pipeline and when the Rotate windows open for each one, it is an identical image that I am cropping each time (i.e. both are the raw images).

I was looking in Windows Photo Viewer, which was the issue. When I use these images in other pipelines for analysis, they appear normal.

Great I will give that a try.
Load_ZVI_Rotate_and_Crop_Pipeline_2.cpproj (139 KB)
ZVI - Processed.zip (7.21 MB)
ZVI - Original.zip (7.21 MB)

[quote=“csirois”]
I used the pipeline that you gave me and am still running into the same issue. I have attached two versions of the same ZVI file - one raw and one subtracted. I ran both of these images through my pipeline and when the Rotate windows open for each one, it is an identical image that I am cropping each time (i.e. both are the raw images).[/quote]

I’m looking at the original and processed images in FIJI (i.e, ImageJ - fiji.sc/Fiji) and they appear to be identical. Are you sure that you are saving what you think you’re saving, or that the background subtraction is actually working…?
-Mark

[quote=“mbray”]I’m looking at the original and processed images in FIJI (i.e, ImageJ - fiji.sc/Fiji) and they appear to be identical. Are you sure that you are saving what you think you’re saving, or that the background subtraction is actually working…?
-Mark[/quote]

When opened in Axiovision, they are different. It seems that whatever modifications are made in Axiovision do not carry over into other software. I guess I’ll have to add modules to my pipeline to subtract the background. Thanks.

Just out of curiosity’s sake, could you post a screenshot of what the original and background-subtracted images should look like?
-Mark

I have attached two screenshots of the image.

I was reading one of the manuals for the Axiovision software and it says:
“If you have modified the display properties of an image you will be asked if you want to save the
changes when you close it, but don’t worry you haven’t really changed the image data. The updated display properties are saved with the image so that it will be displayed in the same way if it is reopened, but the image data has not been changed and the display properties can still be put back to linear.”

I’m thinking that this is the reason why the unaltered version of the image shows up when it is opened with other programs.
Usually I will export the file as a TIF and only use 3 of the 4 channels. This never gives any issues and the processed file can be used by ImageJ, Photoshop, etc. However, I need to use all 4 channels now and a 4 channel TIF seems to throw off most software.