measuring psf without fluorescence

[mete]

I understand that the fluorescent beads are used to measure psf. They are around 200nm in diameter I think. Is there a common method to measure psf without fluorescence ? The minimum pinhole I see is 1um (thorlabs etc.).

[MichaelG.]

Way out of my league

But I’ve just found this

https://www.researchgate.net/post/How_t ... aser_lines

… which might be a good place to start browsing

MichaelG.

[Chas]

I wonder if this has some similarities with the old 'Star test' which I seem to think was done with either pinholes that appear in nigrosin or by bashing globules of mercury with a broad corset stay, for reflected light.
Here is someone using silvered sellotape to make larger holes (not what you are after) but maybe the references might be useful(?)
https://www.quekett.org/wp-content/uplo ... Optics.pdf

[mete]

I wonder if this has some similarities with the old 'Star test' which I seem to think was done with either pinholes that appear in nigrosin or by bashing globules of mercury with a broad corset stay, for reflected light.
Here is someone using silvered sellotape to make larger holes (not what you are after) but maybe the references might be useful(?)
https://www.quekett.org/wp-content/uplo ... Optics.pdf

I think star and similar tests are to assess resolution and some aberrations. The problem is, again I think, they should be above diffraction limit to be useful, whereas to measure psf, something below diffraction limit is needed.

It seems it is quite impossible to find a pinhole smaller than 1um. PCB manufacturing cannot get close to um numbers probably due to etching. Everything else like IC/silicon wafer manufacturing seems to be too advanced for this purpose. Fluorescent beads are really a good idea.

[Chas]

mete, have you looked at the material from a silver helium party balloon?
Here is a ~10 pixel square-ish crop a small spot of light when viewed with a 100x objective on an 18mp APSC camera Jpeg:

Twelfth inch Baker silver helium party balloon crop of tiny hole resampled up to 800.jpg (17.59 KiB) Viewed 3089 times

This balloon was tangled in a hedge so it had lots of larger sized damage to the silvering layer but I wonder if a new one might just have just a few tiny defects to the silvering.
A lot of these ballons seem to have an extra layer of colouring but this was a number-shaped balloon and often these seem to come with just the silvering layer.

[Sure Squintsalot]

I understand that the fluorescent beads are used to measure psf. They are around 200nm in diameter I think. Is there a common method to measure psf without fluorescence ? The minimum pinhole I see is 1um (thorlabs etc.).

Wouldn't a pinhole create diffraction? Which you would not want. And what is a "measurement" of PSF vs. an observation of a PSF? What would be the units of measurement?

[mete]

I understand that the fluorescent beads are used to measure psf. They are around 200nm in diameter I think. Is there a common method to measure psf without fluorescence ? The minimum pinhole I see is 1um (thorlabs etc.).

Wouldn't a pinhole create diffraction? Which you would not want. And what is a "measurement" of PSF vs. an observation of a PSF? What would be the units of measurement?

For deconvolution, ideally the actual PSF of the system has to be known (measured). Measurement means capturing the image of ideally a point source (how image differs from a point) and using this to create/calculate PSF of the system. Practically, as far as I understand, it means to have a source smaller than diffraction limit. That is why ~200nm fluorescent spheres are used.

[Macro_Cosmos]

Well, you can use gold-coated beads for reflected light but I am pretty sure that is not a good answer.

200nm is the typical size of the bead. The size you want to use depends on the system and the method. You will want to use smaller spheres (~100nm) for confocal and superresolution. You may need to use larger ones for lower magnification.
For widefield, I think 175nm is pretty good.

If you want a bootleg solution that does not involve those tiny balls, you can use a projector's reflector mirror and use transmitted light. Adhere a coverslip to the coated side. Imperfections in the coating will reveal themselves as microscopic dots good enough to create airy discs.
(This is known as a star target.)

That said, any measured PSF for convolution is as good as the slide itself. Ideally, you want the slide prepared in the exact same manner as the slide you wish to deconvolve, down to the coverslip thickness and dyes. Otherwise, why even bother? Just do blind deconvolution using Adobe Photoshop.

Here is an example using a couple of star targets I got from a batch of weird Russian parts.

Are these good enough for devonvolution? Nah.

Fluorescence beads are good for a range of examinations, not just PSF. I used them to determine field flatness here:
https://macrocosmosblog.wordpress.com/2 ... decentred/
(Not decentred, the non-plans are just that crooked.)

Just use a theoretical PSF.

[patta]

A rougher version of previous post
Just a black pencil ("Sharpie" ot others) on slide, leaves some tiny holes often suitable for star test.
The metallic mirrors are better though.

[SuiGenerisBrewing]

PSF's are wavelength-dependent, so unless you're imaging with light through a fairly tight notch filter, you're probably not going to be able to measure a PSF very effectively as each wavelength of light passing through your subject will create its own PSF, which will overlap with the PSF of the other wavelengths passing through the same object. Using a theoretical calculator, like this one here, should give you a reasonable estimate (I'd suggest using a wavelength near the middle of human vision, or the highest-intensity emission point of your lightsource if you have its spectra). It may be quite difficult to measure it directly due to the impact of overlapping wavelengths.

For measuring it directly, all you need is something containing sub-resolution particulates. The exact size of those particulates doesn't really matter, so long as they are sub-resolution (which for a high-end oil immersion objective is going to be anything sub-250 um). Tap water often contains small particulate's of this size, although they move so you need pretty fast capture to acquire them. Others have mentioned other useable sources above.

FWIW, the *theoretical* width of a PSF is roughly 0.6λ/NA, so for visible light the width of your PSF can vary by a factor of 1.75 between violet and deep red wavelengths.

[mete]

PSF's are wavelength-dependent, so unless you're imaging with light through a fairly tight notch filter, you're probably not going to be able to measure a PSF very effectively as each wavelength of light passing through your subject will create its own PSF, which will overlap with the PSF of the other wavelengths passing through the same object. Using a theoretical calculator, like this one here, should give you a reasonable estimate (I'd suggest using a wavelength near the middle of human vision, or the highest-intensity emission point of your lightsource if you have its spectra). It may be quite difficult to measure it directly due to the impact of overlapping wavelengths.

Thanks for mentioning wavelength-dependency, I was also wondering this. So it actually makes a lot more sense for fluorescence than for (white-)light microscopy. I already tried psfgenerator and deconvolutionlab2.

[SuiGenerisBrewing]

Thanks for mentioning wavelength-dependency, I was also wondering this. So it actually makes a lot more sense for fluorescence than for (white-)light microscopy. I already tried psfgenerator and deconvolutionlab2.

There still is a PSF with white light, its just that it is not as defined as with fluorescence as you get multiple overlapping PSF's due to the multiple colours (wavelengths) of light present from the light source. Even with fluorescence this occurs to some extent, as even the "best" fluorophores emit a fairly broad spectrum (typically 50-100 nm in width). Are you trying to do white light deconvolution? There are approaches for that, which involve calculating a "phase PSF", which is conceptually similar to a conventional PSF, but for white light. This is then used to deconvolve widefield images via conventional deconvolution algorithms. At least one study has done this in ImageJ, so it is free to give it a try: https://www.researchgate.net/publication/260372545_Direct_Imaging_of_Phase_Objects_Enables_Conventional_Deconvolution_in_Bright_Field_Light_Microscopy

[mete]

Thanks for mentioning wavelength-dependency, I was also wondering this. So it actually makes a lot more sense for fluorescence than for (white-)light microscopy. I already tried psfgenerator and deconvolutionlab2.

There still is a PSF with white light, its just that it is not as defined as with fluorescence as you get multiple overlapping PSF's due to the multiple colours (wavelengths) of light present from the light source. Even with fluorescence this occurs to some extent, as even the "best" fluorophores emit a fairly broad spectrum (typically 50-100 nm in width). Are you trying to do white light deconvolution? There are approaches for that, which involve calculating a "phase PSF", which is conceptually similar to a conventional PSF, but for white light. This is then used to deconvolve widefield images via conventional deconvolution algorithms. At least one study has done this in ImageJ, so it is free to give it a try: https://www.researchgate.net/publication/260372545_Direct_Imaging_of_Phase_Objects_Enables_Conventional_Deconvolution_in_Bright_Field_Light_Microscopy

I think at least theoretically deconvolution microscopy makes a lot of sense and it should not be difficult to apply. That is how I reached this point but I neither have an FL microscope nor have an interest using FL so I wondered if this is applicable to widefield images.

Thanks for the link I will check.