I recently designed and built an epifluorescence macroscope centered around the Olympus XLFLUOR 4x/0.28NA objective and Kinetix camera for both fixed and live imaging. It uses the high power CBT-90 LEDs driven at 27 amps as the excitation source to get up to 690 mW (29 mW/mm^2) of excitation light onto the sample plane for very high temporal resolution imaging.
Specifically, this macroscope has a 4.7 mm x 4.7 mm FOV and the optical resolution is pixel limited with the 10 MP Kinetix camera; having pixels that are equal to 1.5 µm x 1.5 µm at the sample plane. The end result is a nearly 5 mm FOV with subcellular resolution!
If high temporal resolution is not needed, the custom LED source can be replaced with any standard microscope illumination source that uses a 3 mm liquid light guide, and the Kinetix camera could be replaced with an astronomy camera with a sensor that is at least 22 mm x 22 mm, such as this camera: https://www.bhphotovideo.com/c/product/1417197-REG/starlight_xpress_trius_sx_46_usb_hub.html
If you have any questions, please feel free to ask.
Cheers,
Ben Smith
P.S.
The development of this scope and illuminator was funded by the following grant: NIH P30EY003176
Also, definite shout-out to Thorlabs for selling screwball parts such as the 60 mm filter cube and tube lenses with massive apertures, which make this sort of system even possible.
Here is the first test image we got off the scope; this is an image of
endogenous tdTomato fluorescence in a PFA fixed Opn4-Cre/+; floxed
tdTomato/+ P4 mouse retina (link to full resolution raw tiff): Opn4-Cre - tdTom P4 retina.tiff - Google Drive
The image was taken in dynamic range mode (16-bit) with a 50 ms exposure.
The retina wasn’t perfectly flat, so some parts are in focus while others
are a bit out of focus, which is a testament to the high axial resolution
of the 4x objective.
Thanks for sharing this - I know it’s been a few years, but really appreciate the CAD files and the detail in your writeup.
I’ve been trying to reproduce the design and I’m running into some conceptual confusion around the lens train. My assumption going in was that the excitation light would be focused at the back focal plane of the objective (as in a standard epi configuration for Köhler-style illumination), but looking at your layout it seems like the light is entering the objective collimated rather than focused at the BFP. Could you clarify the intent there? Specifically, what motivated that choice, and is the BFP being deliberately bypassed or am I misreading the optical path?
That’s a great question, as you’re right normally you’d want to shine the image of the light source onto the back focal plane of the objective to get Kohler illumination. For this setup, we can leverage the fact that a liquid light guide is basically a low-loss diffuser whose output is a uniform top-hat profile. Therefore, by focusing the image of the LLG onto the sample plane, our illumination is ensured to be a uniform disk of exactly the correct illumination area.
To summarize the light path, the first pair of lenses after the LLG are there to magnify the image of the LLG so that it will fill the FOV of the objective (assuming a FN of 26.5 mm). The biconcave lens then pushes the conjugate focal plane out onto the back focal plane of the objective, and finally the last lens collimates the light.
This four lens setup ensures near perfect coupling of the LLG into the objective. This was paramount as one of our projects is to do widefield voltage imaging, which requires very high illumination powers.
One last note, are you using your own illumination source or the same CBT-90 LEDs we used on our setup? If you are using the CBT-90 LEDs we should set up a time to meet, as the original LED driver used is very painful to build, and there are much simpler options.
Thank you so much for taking the time to walk through this, I really appreciate it. We’re using a Lumencor SOLA coupled to a 3 mm LLG as our source!
I’m still not quite seeing the path though. If collimated light is going into the objective I’d expect a diffraction-limited spot at the sample rather than a wide field, so I’m probably missing something about how the illumination is being configured.
Would you be willing to share the lens spacings or a Zemax file so I can stare at it? That would really help me understand what’s going on.
