Troubles with cell segmentation

[Caterina Tomba]

Hello,

I contact you because I found your software very promising and useful for my project. I'm working on mammalian cells on 2D+ substrates and I would be interested to get information like cell volume and thickness, shape of the apical surface, number of neighbor cells.

I tried to play both with the acquisition parameters (DeltaZ size, number of bits, resolution, scan speed, averaging) and the parameters during segmentation process but the result is far to have a reasonable respect of the real cell morphology.

I'm sorry to ask you your help for probably banal questions, but I would like to know if you consider that it's a problem of staining quality or if it's just that my way to use the software is wrong somewhere.

I will send you some crop images of my acquisitions to give you an idea of what I'm trying to segment. For info, I'm using the Windows version of the software.

Thank you very much for your support. Best regards Caterina Tomba

---

Caterina Tomba, PhD Aurélien Roux lab, Biochemistry Department University of Geneva, Science II quai Ernest Ansermet 30 CH-1211 Geneva 4, Switzerland Mail: [e-mail] Tel: [phone]

[Pierre Barbier de Reuille]

Hey Caterina,

sorry for being so late in answering. But I am catching up with things and should be more responsive from now on. I had a look at the images you sent. I have a few comments:

1. There seems to be a lot of stained elements in your cells. From experience, this can often be a timing issue. But not knowing the stain, I cannot be certain. You should try to reduce their size and relative intensity (e.g. it might be easier to make the membranes brighter than the blobs darker). However, this will cause trouble for the segmentation. There is one process that could help: in the Morphology folder, the process named "Sieve Filter". Its main argument is the size of the objects to remove (when in 3D it will be in cubic micrometers). It will remove all objects at most as big as your threshold, independently from their shape. If the membrane marking is continuous, this should make the membrane big enough that it will be one of the last object removed with this.
2. I am surprised by the absorption of the light while going through the tissue. Even plants often have better penetration than that. Can you increase the laser intensity without damaging your tissue?
3. For the parameters, I would recommend closing the pin-hole to 1 Airy Unit. If the signal is strong enough, you can even go a bit lower than that. Use a Z resolution that will be the same as the X and Y resolutions. This will make a lot of the processing easier, and very often microscopes are good enough that it is useful. Also, if available, use 16 bits acquisition, and don't bother with averaging.

Don't hesitate to ask more questions if you need to.

[Caterina Tomba]

Dear Pierre,

Thank you very much for your reply. Meanwhile, I spent quite a lot of time improving the quality of the images. Unfortunately I tested different concentrations and incubation time but without any significant improvement. Indeed, I used CellMask and SiRActin but the signal was always partially diffused inside the cells too. Finally I recently got new images with a stable cell line labelling the cell membrane. I think the images are now better and I will give you a new feedback as soon as I try to analyse them with your software. Regarding your other questions, I cannot increase the laser also because I get saturated images on the first layers closer to the objective. I already use pinhole 1, no averaging, 16 bits, 1024x1024 size frame and Z resolution of 0.5 or 1um. I remember I also tried to use the Sieve filter but I haven't noticed a significant improvement in the final segmentation. However, I will try to use more systematically.

I will write you soon and I'm obviously very interested to read you again if you have any other suggestions for the software utilisation. Thanks again! Best Caterina

[Pierre Barbier de Reuille]

Hey Caterina, if you have problems with over-saturation, that is when you can go below 1 Airy Unit for the size of the pinhole. The intensity will decrease drastically and you will gain also in resolution. But a more stable labelling is definitely great news!