Problem detecting subcellular object from .nd2 files

Hi. After the webinar on QuPath by RNAscope, I was trying to use the program with my images to detect cells and subcellular objects. While cell detection works perfectly, nothing is detected for subcellular objects. I tried many different settings. My files are .nd2, can it be the problem? I did an analysis with TIFF and it works perfectly.

Could you provide more information about the image type and how it is being read in both cases? The Image tab on the left should have most relevant information.

If you can create cells, then I can’t think of any reason you should not be able to create spots.

Here are the spec for the .tif and .nd2
They both have 3 channels. The .nd2 file is much bigger so I don’t know if it is the size.

1 Like

I suspect the problem is that the ND2 is a Zstack while the TIFF is a small picture with incorrect pixel dimensions :slight_smile:

Try using the slider to go to the bottom of the Z stack (upper left of your image). I suspect you will find the subcellular detections placed there at the bottom of the stack. @petebankhead in case there is any way to get subcellular detections onto the right plane, but I do not think he is a huuuge fan of that particular bit of code :slight_smile:

If you really want to analyze a Zstack with subcellular detections, you may need to do some coding to move them into the correct plane. Or you could try using the Simple Thresholder instead, which I suspect will place nicer with objects in different slices.

If you really want to do 3D analysis though, you might want to look into another program.

Thank you. It’s true that in their workshop, RNAscope used an epifluorescence image, but they mentioned that Zstack should work. I guess it is not the case. We’ve been trying so many ways to quantify the RNAscope images.
I went through each sections of the Zstack and couldn’t find the subcellular detections. I will try the Simple thresholder or I will flatten the stack before uploading it to QuPath. For the moment we are not looking at super precise counting.

1 Like

Many functions will work in Z-stacks, but I do not think subcellular detections has been touched recently. It really requires a bit of scripting to work well with Z-stacks - see:

and resolved issue:

Hi @beamilon and @Research_Associate

This is a bug in QuPath – subcellular detections should work with z-stacks, but currently don’t. I’ve created a bug report here:

I’ve also made some changes to the code that I think should fix the issue here:

These should be included in the next release of QuPath.