PSF Distillation, Depth-Dependent PSFs and PSF Mowdels

Hi @SVI_Huygens Team!

We started using our PSF Distiller in earnest for Lightsheet data and had a few questions coming up, for which we were hoping to get some clarifications from the experts.

  1. PSF Distiller: Is it possible to use a distilled PSF along with Depth Dependent Deconvolution? Inthe case of Lightsheet data, it happens that due to the long working distances and RI mismatches, the PSF varies a lot. Would it somehow be possible to provide multiple distilled PSFs in a depth-dependent manner?
  2. Astigmatic PSFs. Due to the geometry of the sample and the agarose in whic it is embedded, we get very strong astigmatism on our PSFs after a certain depth. Can the theoretical PSF models provided by Huygens handle such cases? In case it is not, it would be quite powerful if we could have this as a parameter, as this has been a very common problem of the LZ1 from Zeiss and any other lightsheet systems that use agarose cylinders.
  3. Regarding Theoretical Depth-Dependent PSF. There is the possibility to work with “some” bricks. How is this selected? How do we know how many are useful or needed? How is the splitting of the data done? Is there overlap? How is the fusion of the different bricks done?

I know this was a rather long post and i thank whoever read it up to here :slight_smile:

We look forward to any and all information you can provide us.




Dear Oli,

Apologies for my late reply I hope you won’t blame Huygens for it. Your questions are very interesting and I needed time to talk with a few colleagues first to collect the right information.

Your questions focus on optimal usage of the PSF Distiller and of the automatic (theoretical) PSF for your Light-sheet data.

Your 1st question was whether it is possible to use a distilled PSF along with Depth Dependent Deconvolution as in practice the PSF changes and varies a lot deeper in your light-sheet samples.

The PSF distiller in Huygens works with the measured beads at one depth at this moment. The reason is that usually beads are imaged at the coverslip. Some years ago Hans van der Voort worked with Hans Gerritsen and Cees J. de Grauw on an article investigating the possibilities to measure beads at specific depths. It was an experimental set-up in which the difficulties in the preparation of the beads at specific depths were the major issue that could not be solved. As far as we know this is still the case but if it could be solved it would open up new possibilities.

The automatic (theoretical) PSF takes into account the shape and differences in thickness of the light-sheet as well as the depth-dependent PSF shapes. The quality of the reported values in the metadata overview by the Light-sheet manufacturers is key, the more correct and more complete the better the automatic, depth-dependent PSF. But as you rightfully mention there are factors of influence that are not included in the metadata.

Also your 2nd question on astigmatic PSFs is a very interesting one.

Setting up a parameter (or more) to capture the depth dependent measure of astigmatism when using agarose cylinders as you propose would still entail a lot of uncertainties and require a great level of knowledge and expertise from the user. Looking at the great possibilities Light-sheet microscopes offer for life-sciences it’s important to keep its use accessible for many scientists.
So my answer circles back to the question whether you or others see possibilities to measure beads at various depths? If so, the article from Hans van der Voort, Hans Gerritsen and Cees J. de Grauw can get follow up and bring new possibilities.

The 3rd question is about the on-the-fly depth dependent PSF measurement. You are absolutely right that there are possibilities to work with bricks. These are selected by Huygens based on the optical parameters and the available GPU RAM. You can best leave the selection on ‘Auto’ as Huygens takes into account all these values and finds the optimal brick-sizes. And there is indeed overlap which is needed for correct fusion or stitching of the various bricks into the deconvolved image.

I hope these answers were helpful. Many thanks for for asking.
Who knows whether your question will trigger experimental research to make samples with depth dependent beads as that is the first step that has to be made.

Kind wishes,

Gitta Hamel

1 Like

Dear Gitta,

Thank you for the very thourough answers!

Regarding producing PSFs at different depths, this is rather simple to do using precisely an agarose cyllinder. We can locate the depth of the PSF easily in the Z stack and acquire many such PSFs so that they cover a wide Z range. One issue there is the presence of the Agarose which will definitely affect the acquisition… But this would actually be the conditions in which the real sample would be anyway.

Concerning the Regarding Theoretical Depth-Dependent PSF, I had an extra question: is there a log somewhere where we can see what sizes are chosen for the brick, the overlaps? Or perhaps some documentation in Huygens I could peruse?

Thanks again and absolutely no worries for the delay. The questions were indeed quite involved :slight_smile:

1 Like