Fluorescence bleedthrough from LED in transmitted light path


I just wanted to post about a pitfall with widefield LED light sources here that I haven’t noticed much discussion around.

The issue is bleedthrough from the widefield illuminiation path into the fluorescence images due to stokes-shifted re-emission from the widefield illumination LED.

What happens:

When taking epi-fluorescence images, some of the fluorescence excitation light travels through the widefield lightpath and hits the widefield illumination LED (if the light path is not shuttered). Due to the phosphors on the LED there will be a stokes-shifted emission that travels back through the transmitted light lightpath and will pass through the emission filter. As a result, one will get an additive mixture of the desired fluorescence image and a transmitted light image. The degree of how strong this is depends on a number of factors (condenser and field stops, wavelength, additional filters) and can range from very subtle to glaringly obvious.

I first noticed this at my previous job and it was not something I was aware of (neither were most of my colleagues). To alleviate this, we placed neutral density and low-pass filters into the transmitted light path. The ND filters will attenuate the light getting to the LED and the reemiited light. If the ND filters are strong enough (we used several), the bleedthrough can be reduced to a level were it becomes insignificant (the LED brightness will have to be cranked up for transmitted light images).
The idea with the low-pass filter was that longer wavelengths may not excite the phosphors in the LED (incoclusive, would need more testing).

The reason I wanted to bring this up and create awareness of it is that I just communicated with someone who returned from a multi-week research visit elsewhere where she spent many weeks acquiring images that have this same issue.

At the facility I work at currently we don’t have a microscope with such a setup, but I’d still be curious to hear about workarounds other than ND filters.


Hey Volker! :wave:

Yeah it’s definitely a thing, and it surprises people a lot.

[A side-note on terminology here: I assume you mean “bright-field” (using the transmitted light path) rather than “widefield” (standard epifluorescence)? If so, can we change the post title to avoid confusion, now that both widefield fluorescence and bright-field often use LED light sources?]

As for workarounds, other than ND filters (and other than simply pushing the transmitted arm back if you’re just imaging in fluorescence):
Some scopes now come with a manual shutter on the transmitted arm that is added specifically for that reason (i.e. it’s not there to shutter the transmitted light). See the image below where I’ve highlighted the shutter on a Nikon Ti2:

Of course, that’s a manual shutter and it won’t really work if you’re looking to take both transmitted light and fluorescence light images of the same sample… (or in a multi-user facility where one user may leave it open and the next may not be aware of it). So to automate blocking of the path, you can also put a blocker in one of the positions of a motorized condenser turret, then program the software to move to the blocked position for all of your fluorescence channels.

That of course assumes you have a motorized condenser turret. If you don’t have a motorized turret, you’re left with either the ND approach, tilting the arm back, or using the built in manual shutter provided on some scopes.


It is so frustrating how few people know about this when it has such a huge impact on imaging! I have found that having a phase ring in the transmitted light condenser does a great job of reducing this (the larger the better–Ph3 is pretty awesome) if you don’t have space for a full block. You can also talk to the light source manufacturer for suggestions on appropriate filters.

1 Like

If you’re not using the transmitted light, putting something to block light (eg, foil or a piece of cardboard) anywhere between the top of the sample and the light source will work too (eg, on top of the condenser turret).


@talley, happy to have the title changed to brightfield or transmitted light.

Unfortunately, most of the workarounds don’t help when one wants to perform sequential acquisition of bright field and epifluorescence.

I found that a McGyver solution when no ND filters are at hand is to put a couple of Kimwipes across the condenser as a makeshift diffusor/ND filter.


Thanks for bringing this important topic up! I also was not aware of these effects. I will distribute this conversation in my area to create more awareness! Best, Josef


Some microscopes allow you to put a shutter in one of the condenser positions. If you have a motorized condenser, you can block the autofluorescence during your fluorescence imaging and open it up for brightfield acquisition.

There are also some transmitted LED light sources that combine 3 non-fluorescent, colored LEDs to make white light rather than use the auto fluorescent white LEDS.

1 Like

The patented bit of cardboard or top from the plastic pot objectives come in - as its black - work for the uprights! Alternatively for fluorescence and brightfield we put a shutter in front of the LED - but we had a spare shutter and a 10-3 and were running things through Micromanager at the time so this was fine. I’m sure on a previous inverted system we had a shutter right in front of the long working distance condensor - I remember it being a nuisance to work with because of the issue of it falling off into an experiment.

Its quite a lot of additional expense though if you haven’t got a spare shutter hanging around and a shutter controller.

Hi - What order are you acquiring the images? We’ve been able to address this when doing phase contrast or brightfield with fluorescence time-lapse imaging by changing the order of acquisition, in this case it was because the phase-contrast lamp needed time to finishing turning off (if that makes sense), and we found if we captured the phase contrast image last, by the time the filters switched (and stage moved) for the next acquisition, we no longer had any phase imaging in our fluorescence channels. Or, as suggested, using or adding a shutter into the transmitted light path should work.

Hi @JamesOrth,
the issue described in my original post is not related to the order of acquisition, the simple presence of a LED light source in the transmitted light path (even if it is never turned on) produces this behaviour.

Ok, I had some trouble interpreting exactly what the issue was, it sounds peculiar, have you had any success determining/addressing the issue?

Yeah it’s a thing @JamesOrth. As @VolkerH mentioned in his first post, it’s due to the phosphors (I’ve been told there’s a phosphorescent screen that the LEDs illuminate…) that is often used in LED illumination systems. The episcopic fluorescence illumination light travels through the condenser lens and into the transmitted light arm, excites the phosphors, and they emit fluorescence that is then collected by the objective lens creating background. You’ll only see it if you have an LED light source instead of a lamp in your transmitted arm.

1 Like

This is a really frustrating issue if you want to combine transmitted light and fluorescence imaging and invested in an LED and a fast camera to do it quickly. The only good solution that doesn’t require hardware to move is the ND filter, which you can add anywhere in the transmitted light arm. You will still get some signal because you don’t block 100% of the light, but it might be too low to be detected with your experimental conditions. Also, you need a reasonably strong LED to overcome the ND filter. If you don’t mind the impact on speed, a 100% block in a motorized condenser (or a shutter) is the safer option.

Very good to know, thanks for the info, I’m glad I didn’t discover this empirically!

This is why I like using green LEDs (that do not have the phosphors mentioned above) for transmitted light rather than white ones (which most often include the phosphors that light up when “epi-illuminated”). Green LEDs are also good for phase and DIC imaging since they are “more monochromatic”. If, for any reason, you still like white LED illumination, there are vendors (I get mine from Sutter) that let you switch out just the LED.

1 Like