Wavelength and resolution (image embedding fixed)

[Wes]

I am sure most if not everyone here is aware of the relationship between wavelength and resolution in a microscope. In essence the shorter the wavelength the better the resolution you get. I decided to test the relation using Frustulia rhomboides, that I got from Klaus Kemp, as a critical test object. Regular white light from an incandescent bulb was used a control against a 447 nm interference filter. The objective I used has a numerical aperture of 0.95 and so does the condenser. Images were captured in monochrome, levels adjusted in photoshop and nothing was changed during image capture except for the addition of the interference filter above the field diaphragm. DIC was used to increase overall contrast.

Frustulia rhomboides has striae period of about 0.29 µm and the resolution I was able to achieve is R=1.22*447nm/(0.95+0.95)=0.29 µm (with some approximation). You can clearly see the improved resolution and detail when using lower wavelengths.

[MichaelG.]

Useful post, Wes ... but I can't see the images

MichaelG.

[Wes]

Yes I just realized. Give me a second to fix the permissions.

[Sauerkraut]

This is very interesting. Thank you for posting.

Intuitively it seems the filter would go between the light source and the objective but could it also be placed between the objective and eye piece? I’m curious about the set up used, having not heard of interference filters before.

Would you see similar improvements in bright field and dark field?

[microb]

Thanks for mentioning the slides source: http://www.diatoms.co.uk/index.htm

I'd like to test various set ups and it looks like this would have some great prepared slides for that.

[MichaelG.]

Yes I just realized. Give me a second to fix the permissions.

I've just woken [it's approaching 06:00hrs here] and seen the images, Wes
... Thanks for fiixing it.
Will study them after topping-up my caffeine level

MichaelG.

[MichaelG.]

... having not heard of interference filters before.

Start here:
https://en.m.wikipedia.org/wiki/Interference_filter

The beauty of them is that they provide near 'brick-wall' slopes.

MichaelG.

[75RR]

Nice test!

Here is a link to a 2008 article by David Walker on Optical interference filters: http://www.microscopy-uk.org.uk/mag/ind ... ilter.html

I don't suppose one can get down to 400nm with a 60w incandescent light source - my impression is that it has to be halogen?

and of course there are the amazing and dangerous UV images at http://www.mikroskopie-ph.de/index.html that show the very limits of light microscopy.

[MichaelG.]

Now that I have some freshly made 'Monsoon Malabar' in my system, Wes
Three brief comments, if I may:

1. A superb demonstration of the phenomenon ... thank you
2. The use of DIC is 'risky' in resolution tests; because it can introduce spurious visual artefacts
3. The SEM with its much shorter wavelength, can show us [to the limits of of our comprehension] what is actually there ... and this is often a useful cross-check of what we appear to be seeing.

https://www.researchgate.net/profile/Pe ... ublication
/233496216_A_morphological_examination_of_Frustulia_Bacillariophyceae_from_the_
Ocala_National_Forest_Florida_USA/links/543fe4ef0cf21227a11b9924.pdf

See Fig.43 et seq.

MichaelG.
.

Edit: The hyperlink is very long, and was upsetting the forum page layout; so I have inserted two line breaks, which will need to be removed when you paste the address.

[75RR]

3. The SEM with its much shorter wavelength, can show us [to the limits of of our comprehension] what is actually there ... and this is often a useful cross-check of what we appear to be seeing.

Apparently SEM images are not immune to artifacts: https://www.sciencelearn.org.nz/videos/ ... in-the-sem

Perhaps UV is more artifact free

[MichaelG.]

Thanks for that reality-check ^^^

It reminds me of:

Big bugs have little bugs upon their backs to bite 'em
... and little bugs have littler bugs,
... and on ad infinitum


MichaelG.

[Hobbyst46]

@Wes: A very nice demonstration.

Intuitively it seems the filter would go between the light source and the objective but could it also be placed between the objective and eye piece?

Isolation of the blue light (447nm) can be realized at any point along the optical path, I think. Yet a filter glass disc, in addition to filtration, refracts and scatters some light, and can ruin the image somewhat if placed at a point which is not designed for it. So it dependes on the microscope. The simplest finite tube length microscope is based on an empty mechanical tube between the objective and eyepiece.

[75RR]

Isolation of the blue light (447nm) can be realized at any point along the optical path, I think

I would have thought it would have to be done before the light reaches the specimen.

[Hobbyst46]

Isolation of the blue light (447nm) can be realized at any point along the optical path, I think

I would have thought it would have to be done before the light reaches the specimen.

My thought, hopefully correct, is that since white light is just a mixture of wavelengths that do not interact with each other, and each wavelength interacts slightly differently with the optical components (for example the Raleigh formula) and the specimen, and since the filter only isolates one wavelength from the others, it does not matter theoretically where the isolation occurs; however, in practice, there are the reflection, refraction, abberations etc caused by each component including the filter (not to mention dust and spots), and this is why the filter is best placed near the light source or adjacent to the field aperture, or near the condenser.

[MichaelG.]

I can offer no actual proof, but: It does seem intuitively obvious that the best resolution will be achieved when [short wavelength] monochromatic light illuminates the specimen, and we use a proper monochrome sensor whose peak sensitivity matches the illumination.

This should minimise the 'noise'

Unfortunately, my mono camera's peak is in the cyan range ... Mmmm

MichaelG.

[MicroBob]

Hi Wes,
thank you for your interesting comparison!

Two remarks:
1. The image taken with the blue light has much more grain. Is this because the filter reduced the light level so the camera used a higher ISO setting? Whey you apply unsharp masking until the white light image is similar grainy - how is the resolution then?

2. Did you use the whole image or select just one color channel? If you took the whole image the blue light picture had the additional benefit that the color error of the objective is taken out of the equation. Can you post the green or blue channel of white light image?

Bob

[Hobbyst46]

I can offer no actual proof, but: It does seem intuitively obvious that the best resolution will be achieved when [short wavelength] monochromatic light illuminates the specimen, and we use a proper monochrome sensor whose peak sensitivity matches the illumination. This should minimise the 'noise' .

Please correct me if I am wrong, but I suggest we should separate the interaction of the wavelength with the specimen, from the intensity of the light source and the sensitivity of the sensor. Fully agreed, that the best would be very intense monochromatic source (ideally, a laser), with a very specifically sensitive sensor (say, a cooled camera or PMT whose peak sensitivity corresponds to the wavelength of the source). The case referred to by Sauerkraut is far from that: an ordinary white light, plus filter, plus ordinary camera sensor. The resolution achieved in both cases is the same. Yet the relative fraction of light wavelength that yields the resolution (out of the total light intensity) is much lower in the second case.

[MichaelG.]

I can offer no actual proof, but: It does seem intuitively obvious that the best resolution will be achieved when [short wavelength] monochromatic light illuminates the specimen, and we use a proper monochrome sensor whose peak sensitivity matches the illumination. This should minimise the 'noise' .

Please correct me if I am wrong, but I suggest we should separate the interaction of the wavelength with the specimen, from the intensity of the light source and the sensitivity of the sensor. [ ... ]

I cannot correct you ... because I think you are right
But, I don't really think we are in disagreement anyway

I struggled to find a better word than 'resolution' to describe the combined effects of the separate factors that you mention ... but I could not.
Resolution is said to occur where we have the ability to just distinguish two individual features; and in practical terms, anything and everything that reduces that ability is effectively reducing resolution.

That said: For the methodical analysis of why we achieve a particular resolving power, it would be helpful to treat each of the contributing factors separately.

MichaelG.

.

Edit: I think perhaps the problem is that I wrote in the context of what Wes was demonstrating ... which is resolution as presented in a final image [being the output from a whole chain of optical and electronic factors]. The resolution provided by the objective is, of course, modified in some way by every additional stage in the process.
.

[Roldorf]

It would be interesting to see what a stack of the different wavelengths of light would turn out. Would it look the same as the white light image?

[Hobbyst46]

It would be interesting to see what a stack of the different wavelengths of light would turn out. Would it look the same as the white light image?

The images you posted nicely demonstrate the advantage of short wavelength for resolution. Dealing with a "stack" is perhaps practical by following the route that MicroBob suggested above: separate analysis of each of the three color channels (R, G, B). However, to achieve a quantitative conclusion, visual inspection of the image would be inadequate.

[MichaelG.]

It would be interesting to see ...

The images you posted nicely demonstrate ...

Forgive me, please, if I have missed something, but to which of Roldorf's images do you refer ?

MichaelG.
.

Edit, just for clarity: My question was intended for, and has been answered by, Hobbyst46

[Roldorf]

Hi MichaelG.
I meant that if Wes was to take images of the diatom with red and green filters then stack them would they end up with the same apparent resolution as the white light image. That was what Hobbyist46 was referring to.

[Wes]

Thanks everyone for the lively discussion, I'm certainly getting new ideas from reading your posts!

1. The image taken with the blue light has much more grain. Is this because the filter reduced the light level so the camera used a higher ISO setting? Whey you apply unsharp masking until the white light image is similar grainy - how is the resolution then?

The ISO is the same for both images. Because the filter limits the total amount of light reaching the sensor the original image is rather dark. I changed the levels in photoshop to have the two images in roughly similar brightness and this is where the graininess comes from. When I apply unsharp mask the white light image becomes a little better but there are still plenty of areas where the pores are unresolved, its the ones that are visible become very pronounced. It has to be kept in mind that the white light contains all of the blue part of the spectrum but since all other wavelengths are also present the image quality deteriorates due to interference (but I may be wrong here so feel free to correct me). What I could do is focus on the pores and try green and red interference filters, that might be informative.

2. Did you use the whole image or select just one color channel? If you took the whole image the blue light picture had the additional benefit that the color error of the objective is taken out of the equation. Can you post the green or blue channel of white light image?

I set the camera to monochrome mode so individually changing the RGB levels in photoshop makes no visible difference in the resolution. I see your point though and I will take a regular image next time to see the individual contributions of the red, green and blue channels.

[Hobbyst46]

It would be interesting to see ...

The images you posted nicely demonstrate ...

Forgive me, please, if I have missed something, but to which of Roldorf's images do you refer ?

MichaelG.

I am sorry for the stupid mistake!! I meant the blue vs white of Frustulia posted above by Wes.

[apochronaut]

This is an excellent demonstration.
One recurring theme on this forum is the one of whether the condenser limits the objective N.A. to the condenser N.A. There are still adherents to this position.
It would be easy to establish a similar comparison where only the condenser N.A. was the variable with a high N.A. dry condenser matched to the same N.A. objective as a control.

[MichaelG.]

... I meant the blue vs white of Frustulia posted above by Wes.

That's fine ... I'm with you now
Thanks for the clarification.

MichaelG.

[Sauerkraut]

Here is a link to a 2008 article by David Walker on Optical interference filters: http://www.microscopy-uk.org.uk/mag/ind ... ilter.html

Thanks for this link. Clever how the excess field iris light was blocked out to avoid reflections using a card stop.

According to the article these filters would be for camera use only to avoid harm to vision(?). Yet I'm still wondering if this technique would improve images on the whole (bright field, dark field), or just for oblique, DIC, etc. methods. Could be interesting to experiment.

[75RR]

https://www.researchgate.net/profile/Pe ... ublication
/233496216_A_morphological_examination_of_Frustulia_Bacillariophyceae_from_the_
Ocala_National_Forest_Florida_USA/links/543fe4ef0cf21227a11b9924.pdf

See Fig.43 et seq.

MichaelG.

Edit: :oops: The hyperlink is very long, and was upsetting the forum page layout; so I have inserted two line breaks, which will need to be removed when you paste the address.

Not sure why the address you have is that long - here is a shorter one that links to the same article:

https://www.researchgate.net/publicatio ... lorida_USA

[MicroBob]

I set the camera to monochrome mode so individually changing the RGB levels in photoshop makes no visible difference in the resolution. I see your point though and I will take a regular image next time to see the individual contributions of the red, green and blue channels.

For diatoms I have done this occasionally and the resulting image of one colour channel changed to b/w was somewhat better than all colour channels together. The method makes an apochromat from a simple achromat in a way.

Here two images, one where I converted the full image to b/w, one is only the green channel (best) converted to b/w. There are probably objects that profit more. This is more or less from the center of the image, so the colour correction effect isn't so strong anyway.

Bob

[MichaelG.]

Not sure why the address you have is that long - here is a shorter one that links to the same article:

https://www.researchgate.net/publicatio ... lorida_USA

It may be because I am using the iPad and/or because of the numerous underscores in the address.

All I did was copy and paste the URL from the web page ...

Thanks for posting the shortened form of the link.
How do you make it do that ?

MichaelG.