Hello everyone,
I’m a biologist and new to this forum. I’ve encountered a microscopy challenge and would appreciate any ideas to tackling it.
I need to image bacterial colonies that grow across the curved surface of U-bottom polystyrene wells in 96-well plates. I’d like to obtain an overview image of the wells to map colony morphology, that is, to differentiate features in the size range of hundreds of micrometer. There’s no need for high magnification optics. Obviously, since growth does not occur on a flat plane in this setting, regular microscopy objectives or macrophotography lenses produce images with only a segment of the image in-focus. Unfortunately, for this experiment it is not an option to use regular flat-bottom wells.
I have experimented with automated scanning wells using a 4x microscopy objective, capturing image stacks in Z direction. The image stacks can be turned into an in-focus image using focus stacking. These can then be stitched to create an overview image of the well. This produces satisfactory results with regards to image resolution. However, this procedure is quite time-consuming, both in obtaining X-Y-Z images, and in manual focus stacking - so far I’ve only found consumer-type applications like Photoshop or Zerenestacker for this purpose. Ideally I’d need to scale this to many wells and experiments, so I’m looking for an optimized solution.
I have considered using a lower-magnification, macrophotography objectives, which would decrease the number of images required to stitch to obtain a whole-well image view. However, the depth of field would remain shallow, probably still requiring focus stacking. I have thought about decreasing aperture with a diaphragm for (possibly?) increasing depth of field, or trying to match the field curvature of (non-plan) objectives with the curvature of the wells (perhaps using wide-angle lenses?). However, this type of hypotheses are quite far from my comfort zone.
I would greatly appreciate hearing from anyone who has experience with imaging in this type of plates or carriers or who would have any other ideas to solve this problem.
Many thanks in advance!
Best regards,
Klaas
1 Like
I don’t have experience in imaging wells however I do have experience in automating microscopy imaging procedures like stereology with .NET applications, finding the top & bottom for taking Z-depth image stacks with a confocal. First taking a whole slide image in 4X then taking z-stacks with 100x based on a stereological procedure. What kind of microscope do you have? I have made programs for Zeiss, Prior, and Olympus. I have made a microscopy imaging application that works on various microscopes called BioImager. Which can be scripted in C# to automate various imaging procedures.
1 Like
Can you clarify regarding the minimum x/y resolution that you need for your application? 10um? 100um? This will determine the minimum NA that you need to take the images you want, as well as what your depth of field will be. If it is 100um it seems like a macro photography lens mounted on a DSLR or equivalent camera would be the way to go. I’d imagine that such a lens would also have a FOV that would cover multiple wells, so I don’t think imaging should take that long even though it may still require some stacking?
Are you limited to trying to do this using a upright and/or inverted microscope? Do you need to take the pictures from above or below the wells? If you are taking the images from below the wells I’d guess that the FOV of most non-plan objectives would curve in the opposite direction from the well bottom so that probably wouldn’t help you much.
Are the wells deep such that the bottom could be occluded by a wall depending on the lens location?
-Hazen
1 Like
Hello Erik, thank you for suggesting BioImager. Seems like a useful solution for sophisticated scanning routines. However, I’m currently using a custom optical setup with custom motor drivers, so I rely on my own scripts for scanning.
Thank you Hazen for your questions and insights. The XY resolution should be around 100 µm. I can mount an imaging unit both in inverted and upright position. I agree that to match the field curvature I should image from above (upright configuration). I found this discussion on photo stackexchange, which suggests that for simple lenses the shape of the field curvature matches the field of view (an uncorrected lens with a FOV of 45° will have a field curvature the shape of a 45° arc of a sphere).
I’ll be using a Sony IMX477 sensor (from the Raspberry Pi High Quality Camera) for prototyping.
It has a diagonal sensor size of 7.9 mm. The wells have a horizontal diameter of 7 mm, let’s assume they are perfect half spheres.
So the field of view of a (presumably) uncorrected f = 4 mm lens would be:
FOV = 2 * arctan(7.9 mm / (2 * 4 mm))
FOV = 89.3 °
Such a field of view of around 90° seems appropriate for imaging half of the half-sphere shape of U-bottom wells, and the 90° arc shape of the field curvature would match the curved bottom of the well?
In the upright position, the lens is above the lid of the plate, at a distance of around 14 mm from the bottom of the well. I’m not sure where the ‘working distance’ of at least 14 mm comes into play here?
All this seems a bit like a ‘bric-a-brac’ solution, so I’d be happy to hear if anyone would take this in another direction.
I think that if you have a system with a very small NA your depth of field will be such that the entire well bottom is more or less in focus. Based on your numbers, for a 7mm diameter sphere you’d need a depth of field of 3.5mm. Referring to this NikonU page:
The Diffraction Barrier in Optical Microscopy
A system with a NA of about 0.017 would have a maximum Z resolution of 3.5mm and a X/Y resolution of about 15um. Desired NA = sqrt(2x0.5um/3500um). XYr = 0.5um/(2.0x0.017) = 14.8um. 500nm (0.5um) wavelength light. An example system would be a 10cm lens with a 1.6mm aperture. I’m assuming that you are doing brightfield imaging here, so illumination shouldn’t be too much of a problem given the very low detection that such a system would have.
Another direction, if you really want to challenge yourself, is a light field setup.
Light Field Camera
If you are at/near Georgia Tech, Shu Jia has been doing some interesting work light field imaging.
Fourier light-field microscopy
-Hazen
1 Like
Thanks Hazen. Intriguing technology - light-field microscopy, but out of reach for me I’m afraid.
Thanks for you suggestion of a 100 mm lens with 1.6 mm aperture - I’ll be experimenting.