Example data for advanced workflow

[tischi]

@ssgpers @manerotoni @seadaisy

Do you happen have example images, which are:

• 2D
• There is a treatment and a control image with a clear difference in the number and/or shape of objects
• One needs to apply a simple denoising (e.g. median filter) for the binarization to work.

Along the lines of something like this (ideally with a bit more objects and ideally some published data).

[manerotoni]

What is the problem with this image? Apart that it has only 2 objects. I don't think this kind of data must come from a publication.

[tischi]

What is the problem with this image? Apart that it has only 2 objects. I don't think this kind of data must come from a publication.

That's the example image that I use in the module to explain, e.g., a median filter. I thought for the workflow we might like to have something else.

[tischi]

I think for the advanced workflow also something where we need a bit of object filtering, right?

[tischi]

It is incredible how much time it takes to find good example data with just the problem that you want ;-) I ended up "manufacturing" two images, by cropping and adding some noise to raw data that we acquired during a course.

The task would be to develop a workflow that automatically measures that the nuclei on the left are bigger than on the right.

[manerotoni]

Sorry I misunderstood the first question. I thought you wanted an additional image for just median filter explanation. For the workflow shape measurement with filters this looks good. I realized that I have also more examples with PLK1 inhibition. I guess I also have to add some noise :-( .
In principlesite does not harm to provide several workflow example images.
The PLK1 difference is in the morphology and can be quite nicely shown with circularity.

[tischi]

In principle site does not harm to provide several workflow example images.

Yes, please go ahead and add more!

[seadaisy]

Hi both, If needed, we have a plate with fixed cells with PLK1 and KIF11 phenotypes (cells have a nuclear marker and labeled tubulin). I could take images in the condition you need for your teaching. All the best Sabine

[manerotoni]

@tischi, and @seadaisy, I uploaded PLK1 example images (xy_8bit__nuclei_PLK1_control, + inhibition). I think they are noisy enough that a simple threshold gives a rough boundary so either one needs raw image filtering or binary operations to clean up.

If @seadaisy wants to acquire some additional data it could be nice. For instance the KiF11 one could also look at the tubulin phenotype (intensity values). In principle one would need a 2D image (quite small 500-600 pixels) with not too many nuclei so that we can easily separate them. One can acquire images at different quality (one rather low but still visible and a good image in both channels). For an LSM for instance by reducing the line-averaging, laser-power, or detector gain. Thanks

Antonio

[manerotoni]

[seadaisy]

f @seadaisy wants to acquire some additional data it could be nice. For instance the KiF11 one could also look at the tubulin phenotype (intensity values). In principle one would need a 2D image (quite small 500-600 pixels) with not too many nuclei so that we can easily separate them. One can acquire images at different quality (one rather low but still visible and a good image in both channels). For an LSM for instance by reducing the line-averaging, laser-power, or detector gain. Thanks

Does it has to be a confocal image? If not, I can do it on a wide- field (very low light intensity should work). Sabine

[manerotoni]

Wide field is fine. They tend to be quite less noisy.

[seadaisy]

Dear All, unfortunately the plate I was talking about is not suitable anymore :( When ever we will have such a plate again, I will come back to you. Nonetheless I took images were always a pro-, or pro-meta-phase is in a field of view together with inter-phase nuclei with high and low exposure time (tubulin GFP and H2B mcherry) if this would be helpful, I could share the images. best Sabine

[tischi]

@imagejan @ssgpers @manerotoni @seadaisy What would be super nice if someone had images like the ones below but in 3D!

[seadaisy]

Shouldn't we maybe collect a "wish list" which data /datasets are needed and try to acquire them? We could then even store them, were the community has access to this data. It needs a bit of planning, but we have the microscopes, I could prepare the samples... What do you think?

Sabine

[tischi]

Shouldn't we maybe collect a "wish list" which data /datasets are needed and try to acquire them? We could then even store them, were the community has access to this data. It needs a bit of planning, but we have the microscopes, I could prepare the samples... What do you think?

❤️ Amazing! Thank you so much for offering this! I think @manerotoni also would have a chance to contribute. I started a google doc for this: https://docs.google.com/document/d/1AkP1Vok5SOcj-eKAR397kpN2XlXXAAjBas8Odoke57I/edit?usp=sharing