Multimodale Koregistrierung an einem Standardmikroskop

[Ramon]

Hello Hubert

It is clear that a blue LED also emits longer-wavelength light tending towards the green. My graph shows the emission spectrum from the datasheet of my Nichia LED, which has a peak wavelength of 463 nm; this applies similarly to any LED. As the fluorescence efficiency is often only 1 per cent, or even less, unless special fluorescent dyes are used, any external light of this magnitude masks the fluorescence.

The UV LED also has a similar emission spectrum of comparable relative width, but I think this is barely relevant in the long-wavelength range >500 nm when the emission peak lies 100 nm away.

Duly noted. This green leakage alters many of my interpretations of different samples under FrTIR fluorescence.

My question about the exposure time refers to the fact that a very long exposure time could be affected by stray light from the room lighting. Your lighting setup also picks up ambient light; if the room isn't darkened, this can also become a factor. I would carry out a blank test without LED lighting using the same exposure time.

I carried the blank test, it came out negative (see below) but I think that your point is fair. I live on the sunny Mediterranean coast, where we get a lot of bright days.



The shading in the light guide comes from the fact that these guides are designed for light entering as a collimated or convergent beam, which requires discrete LEDs and lenses. For me it's much cheaper to use LED strips (a wide beam), which allow for simpler light guides. But with a wide beam, some of the light breaks the TIR condition before reaching the sample and ruins the contrast — hence the shading.

Bear in mind that these light guides are CNC-machined, and the cost rises sharply with complex geometries, such as the recesses needed to fit the lenses. Below is a photo of the two types of light guide I use: on the left, the discrete-LEDs-and-lenses type; on the right, the LED-strip type. It's also worth mentioning that, for proper contrast, the discrete-LEDs + lenses type needs to have its perimeter blackened.

It's clear that this is scattered light just because there are no fluorescent plant structures there. Perhaps there are scattering particles in the embedding medium of the specimen. These could be illuminated by emissions scattered horizontally from the nearby specimen. I consider this commercial specimen unsuitable for conducting fluorescence experiments. Your own plant section also shows – as far as can be seen in the image – no scattered light outside the section.

I agree. Over the past two days I've been thinking about all kinds of possible explanations. One was that the mounting medium was fluorescing (mounting resins are known to fluoresce yellow-green), but I ran a test with Canada balsam and its fluorescence was far too faint to account for the signal.

Please let me know whether you agree with the following interpretation of the image below (Marchantia section with antheridia at 5× from a prepared sample, under brightfield, FrTIR/White, and FrTIR/Blue + Yellow LPF): what we're seeing in the FrTIR/B+Y LPF image is the green leaked from the blue LEDs and scattered by the cell structures (possibly with some added yellow-green from the yellow long-pass filter itself, which is a cheap one that can self-fluoresce) — except, perhaps, for some genuine fluorescence in the greener structure on the right side of the image.



And thank you, sincerely, for the time you spend on these answers — they're consistently informative, and precisely the kind of feedback I was hoping to find when I joined this community.

Ramon, from sunny Catalonia Spain 

[Lupus]

Hello Ramon,

As has already been discussed, combining blue light excitation with a 500 nm cut-off filter – which still allows a significant proportion of the longer-wavelength excitation light to pass through – is not suitable for a clear demonstration of fluorescence.

In the image below, I have provided a comparison between dark-field illumination using white light (left) and blue light excitation (right). You can see individual structures that glow significantly brighter on the left and are therefore exhibiting fluorescence.
The image intensity has been adjusted in some pictures through image processing.



I can barely detect any difference in fluorescence between green and red emissions, except in small areas at the bottom right. I am not a biologist; with the relevant knowledge, one could explain which areas should theoretically exhibit fluorescence due to the plant's structure. There is a good reason why one always uses a suitable set of narrow-band excitation and cut-off filters in fluorescence microscopy. This is also because fluorescence often occurs relatively close to the excitation wavelength.

In order to distinguish between red chlorophyll fluorescence and the green fluorescence of other cellular components, one would need to take one image using a blue LED and a red cut-off filter, and another using a UV LED and a green bandpass filter.

Hubert

[Ramon]

Hi Hubert,

Well, I think you explained it all. I am comfortable with all you said.

Following up on your point that residual excitation still registers in the G and (less) the R channel — and, bigger still, the green the LEDs leak straight into the filter's passband — I've asked the AI to check, pixel by pixel, whether the red channel is simply a weighted combination of the green and blue channels. I've now put a number on it, on the images already in this thread (the winter-jasmine and the Marchantia) plus one clean control I ran to isolate the effect. Short version: the "red" in these images, under blue/violet excitation with no red in the illumination, is a camera artifact — the red channel is almost entirely reconstructable from the green and blue channels.

A word on the notation, since it recurs below. For each pixel I write its red value as a weighted combination of its own green and blue values plus a constant — R = a·G + b·B + c — and let a least-squares fit choose the weights (a, b) and the offset (c) that best reproduce the actual red across the whole image. The R² that comes with it scores the match: 0 means no relation, 1 means the red is reproduced perfectly. If the red is nothing but crosstalk from green and blue, this simple formula fits it well and R² sits near 1. (The offset c is small, so I drop it from the fits quoted below and report only the weights a and b.)

1. The winter-jasmine. The fit on my blue + yellow-LPF capture: R ≈ 0.26·G + 0.89·B (R² = 0.89) — ~90% of the red reconstructed from green and blue. So the "red areas" that looked like they might be chlorophyll are crosstalk.

2. The Marchantia. Same result, and at the same strength as the jasmine. On a full-RGB capture of the section: R ≈ 0.28·G + 0.66·B (R² = 0.87) under blue excitation, and R² = 0.91 under UV. Your DF separation already showed the red tracking the green; with the blue channel available too, the full two-term fit confirms it — the red is a combination of green and blue, nothing more.

3. A clean control (new) — a moss section. The moss is the cleanest of the three: green and blue are both strongly present, so the two terms separate well and the fit is tightest:

R ≈ 0.39·G + 0.63·B  (R² = 0.96)

96% of the red channel reconstructed from green and blue — the cleanest demonstration that the red here is simply a recipe of the other two, not a record of red light.



Channel separation:



The three fits together — every one of them says the red channel is a linear combination of green and blue:



A methodological corollary, following your own point. Because the red is reconstructable as R ≈ a·G + b·B, amplifying the red channel to look for chlorophyll doesn't really test for it: brightening the red just brightens a scaled copy of green and blue, so "red areas" appear whether or not any chlorophyll is present — and the amplification makes that artifact look like real structure. The honest test isn't the amplified red image but whether any red survives above the a·G + b·B prediction (the "excess"). In these images almost none does — which is exactly what the R² approaching unity reflects.

Honest caveat / next step. The real fix is upstream — an excitation band-pass filter to strip the LEDs' green (and residual blue) at the source, which is what the FrTIR gadget still lacks. You had already flagged this gap yourself.

I've been sourcing excitation filters to place over the LEDs, and the Rosco and LEE lighting gels look like the most affordable fit. The one caution I've had is that such gels cut the excitation intensity a lot — which, with an already faint fluorescence, could leave the signal too dim to capture.

Thank you again — this exchange has been highly informative.

Ramon

[reblaus]

Hi Ramon -

The fluorescence phenomena you're anaylzing are not only due to a mixture of single components but by their interaction in a very complex photosynthetic system. That makes it very difficult to interpret your observations.
    Around 50 years ago we have been studying delayed fluorescence of plant leaves caused by inhibitors of photosynthesis like draught or herbicides - e.g. Simazin which inhibits the transfer of electrons from the excited chlorophyll molecules. The unused energy is then returned by a delayed red fluorescense (sort of luminescence).
    In this context  it  would be interesting to know whether the photosynthetic system in your fresh preps is still intact.
See the following article in Wikipedia:

https://en.wikipedia.org/wiki/Chlorophyll_fluorescence

Good luck with your illumination device!

Rolf

[Ramon]

Hello Rolf

The fluorescence phenomena you're anaylzing are not only due to a mixture of single components but by their interaction in a very complex photosynthetic system. That makes it very difficult to interpret your observations.
    Around 50 years ago we have been studying delayed fluorescence of plant leaves caused by inhibitors of photosynthesis like draught or herbicides - e.g. Simazin which inhibits the transfer of electrons from the excited chlorophyll molecules. The unused energy is then returned by a delayed red fluorescense (sort of luminescence).
    In this context  it  would be interesting to know whether the photosynthetic system in your fresh preps is still intact.
See the following article in Wikipedia:

https://en.wikipedia.org/wiki/Chlorophyll_fluorescence

Good luck with your illumination device!

I am starting to find out that fluorescence mikroscopy is not a joke. The more I learn the more I realize that there is so much that I dont know that I feel humbled. I find it beautiful and fascinating. Few weeks ago I knew nothing about fluorescence. I just aim at being able to do some kind of amateurish fluorescence that can interest entry level microscopists. I will keep posting my journey into the interesting field of fluorescence mikroscopy and other applications of the gadget.

Thanks a lot for your warm ending words 

Ramon

[Lupus]

Hello Ramon,

Here is a short post on autofluorescence, which includes two good examples of simultaneous green fluorescence from lignified cell regions and red chlorophyll fluorescence.
https://www.mikroskopie-bonn.de/bibliothek/botanik/151.html

One of the authors has also posted a few threads on this topic here in the forum, here is an example:
https://www.mikroskopie-forum.de/index.php?topic=35600.msg260410;topicseen#msg260410

However, he always used high-quality Zeiss filter sets, which allow the excitation and emission wavelengths to be closer together.
And the plant sections were freshly prepared in each case.

Hubert

[Ramon]

Hello Hubert and Rolf,

Finally the LEE excitation filters arrived. I repeated the test using an excitation filter to remove the green leak of the blue LEDs. And the results are not what I expected.

In our earlier exchange we concluded that the green "halo" around the jasmine section was most likely scattered green leaking from my blue LEDs, rather than fluorescence. I have since built a simple DVD-grating spectroscope and run a test that disputes what we thought was clear.

The LED spectrum and the gel (Fig. 1 — qualitative)

•   White LED: full continuous spectrum — reference.
•   My blue LED: a dominant blue peak, but with a clear green tail (leak) and a faint red — so the LED does leak green (and a little red), as suspected.
•   Blue LED + a LEE "Deep Blue" 120 gel over the source: the green and red are gone; essentially clean blue reaches the sample.

This is a DIY spectroscope + phone camera, so please treat it as qualitative, not calibrated — but the leak and its removal by the gel are unambiguous. LEE's own published transmission curve for this gel (Fig. 1b) confirms it quantitatively: it passes ~38% at my 450 nm excitation line but drops the green-leak band to <0.3%. (Its quoted Y = 1.1–2.1% is luminous, i.e. eye-weighted, transmission — which is why the gel looks almost opaque yet still lets the blue excitation through.)


Fig. 1 — DIY DVD-grating spectroscope. White LED (full spectrum), blue LED (blue peak + green/red leak), blue LED + Deep Blue 120 (leak removed). Qualitative; both blue panels brightened equally.


Fig. 1b — LEE 120 Deep Blue spectral transmission, plotted from LEE Filters' published data. ~38% at 450 nm; green-leak band <0.3%.

The prediction.

If the halo is scattered leaked-green, then filtering that green out at the source (Deep Blue 120) should make the halo disappear.

The result (Fig. 2).

It does not. Same section, same blue excitation, same Thorlabs 500 nm emission filter — the only change is the Deep Blue 120 gel on the source. The green halo is still clearly there. A rough halo-to-tissue green ratio (which cancels the exposure difference between the two frames — they were matched by eye at the eyepiece, not metered) is essentially unchanged: ~0.40 without the gel, ~0.42 with it.


Fig. 2 — Same jasmine section, blue excitation, Thorlabs 500 nm emission. Only the Deep Blue 120 excitation gel differs. The green halo persists.

Removing the green leak does not remove the halo, so "scattered leaked green" explanation must be put on hold.

Honestly guys, this is as far as I can go. I have no explanation. Help is very welcome here! 

Ramon

[Lupus]

Hi Ramon,

I'm not familiar with plant sections, but if I understand correctly, you are using an older, commercially available, and stained section of jasmine. In my opinion, that is an unsuitable starting point; the strong scattered light between the cell walls is also a clue.

My fresh cutting in water (post #17) was also excited by a blue LED. The green here is not fluorescence, it is also blue excitation of the green color channel - but mainly on the cell walls. I doubt that green fluorescence can be observed with blue excitation using such simple tools.

Hubert

[reblaus]

Hi Ramon,

I've got quite a number of stem cuttings embedded in Euparal showing similar effects. I think they are due to diffusion of stains since it wasn' present in fresh preps but increased during time. I suppose some of the compounds are soluble in the Euparal solvent (Isoprop ?)

Rolf

[Ramon]

My fresh cutting in water (post #17) was also excited by a blue LED. The green here is not fluorescence, it is also blue excitation of the green color channel - but mainly on the cell walls. I doubt that green fluorescence can be observed with blue excitation using such simple tools.

Hi Hubert

On the jasmine — you're right. It's a commercial slide bought online (unknown staining and mounting), a poor starting point exactly as you say. Let's forget about it.

Moving on, your opinion regarding the green fluorescence on the FrTIR gadget leaves me in a quite uncomfortable spot. Let me explain. 

I my post #4 regarding the green fluorescence on the fern leaf I explain that this very sample was sent to Pau (Photomacrography.net) — whose expertise I trust — he imaged it on a professional epifluorescence microscope and read it as genuine green fluorescence, and the FrTIR gadget reproduces the same green cell-wall pattern. So I am holding two expert readings (you and Pau) that point in opposite directions, and I don't feel I can honestly claim fluorescence until I understand why. I'd genuinely like to come clear on this with you: how do you read the fern result?

Regards
Ramon

[Lupus]

Hi Ramon,

I don't see any contradiction in the opinions; my doubt concerned the winter jasmine. This specimen apparently exhibits strong scattered light, whereas the fern leaf does not. The fern was excited with violet light (400 nm). My most recent doubt ("blue excitation using simple means") referred to the close proximity of the blue excitation spectrum - without a blocking filter - to the 500 nm long-pass filter.

Hubert

[Ramon]

I've got quite a number of stem cuttings embedded in Euparal showing similar effects. I think they are due to diffusion of stains since it wasn' present in fresh preps but increased during time. I suppose some of the compounds are soluble in the Euparal solvent (Isoprop ?)

Hi Rolf

That makes a lot of sense. Do you think that these stains are fluorescing or scattering?

Ramon

[reblaus]

Hi Ramon -

It might be both. Looking for a wellknown substance contained in the cuttings (stilbene, blue fluorescence) I had no time to worry about side effects which didn'disturb at that time.
But I'll see whether I can find the old slides.

Rolf

[Ramon]

I don't see any contradiction in the opinions; my doubt concerned the winter jasmine. This specimen apparently exhibits strong scattered light, whereas the fern leaf does not. The fern was excited with violet light (400 nm). My most recent doubt ("blue excitation using simple means") referred to the close proximity of the blue excitation spectrum - without a blocking filter - to the 500 nm long-pass filter.

Hi Hubert. My apologies, I miss read your post. And honestly, I'm relieved! It's good to know you and Pau are essentially of the same mind on the fern.

On the optics: a 450 nm excitation with a 500 nm long-pass sits too close together — too small a separation for a clean test of green fluorescence. A 400 nm excitation with the same 500 nm cutoff is the cleaner pairing. Now I get it.

Regarding the green halo on the Jasmine. Could it be scattering and fluorescence as Rolf suggest?

Ramon

[Ramon]

It might be both. Looking for a wellknown substance contained in the cuttings (stilbene, blue fluorescence) I had no time to worry about side effects which didn'disturb at that time.
But I'll see whether I can find the old slides.

Hi Rolf. Perfect!

Big thanks
Ramon