I am planning to build a system for monitoring pollen in the air. I have made a small gadget to collect pollen from the air and I am in the process of making the OFM. The last piece will be some software to recognize the pollen in the microscope images. Anyone else having similar ideas/done something similar?
I’ve definitely spoken to folk interested in measuring PM2.5 with the OFM (Public Lab for example) but I’m not sure how far they got with automated analysis.
Great, appreciate if you notice me if they get in touch again. One good thing about pollen is that it is much larger than PM2.5, pollen is 25-60 um, and 100x is the normal magnification used. Which should make it possible to see it with OFM.
sure - I’d expect you would see the pollen in excruciating detail at 100x, if I remember correctly the field of view is only about 150um across at that magnification. If you’re counting grains (rather than examining one grain in close detail) I’d have though a lower magnification would make sense so you can fit more of them in each image. However, I’ve not done pollen counting so perhaps something’s going on that I am not aware of. It is, however, worth checking that when I say 100x I mean a 100x objective, if you are comparing to a 100x total magnification with 10x eyepieces, that’s what I’d call 10x. At that magnification I’d still expect you could see them very well, but you’d fit a lot more in each image…
I am looking at pollen with a regular microscope, using 10x objective and 10x in the ocular. This is OK in terms of seeing the details. I was assuming that 100x objective in OFM would give the same magnification, but I am an amateur and might have got this wrong.
My understanding is that 100x would give a field of view that is 2x2mm, e.g. the same as the movement with the step motors. The sample plate I am using is 110x110 mm, so I am planning to patch together 6x6 images to make a picture of the whole plate, before I run that through the image recognition software.
There is a document somewhere on the openflexure web site with field of view for different objectives. I’ll edit it in here if I find it.
Edit: It is on the Wiki:
That is the old location - I can’t find the new one just now!
110mm is 60x60 images at 2mm square, and would need more for overlap for tiling. At this size the 2mm motorized x-y motion is not much use for you. Maybe the flat top microscope would be better?
Apologies, the sample plate is 11x11 mm. But it might still be difficult? The flat top looks very cool btw
Ah ok, glad I asked! This is a recurring issue, because some people are used to quoting ocular x objective, and others just talk about the objective. In the OFM, there are no eyepeices, but the small digital sensor and larger screen mean that the effective ocular magnification is something like 10-20x. So if you’re using a 10x objective at the moment, I’d use a 10x objective with the OFM.
It’s worth pointing out, though, that the travel of the stage is only about 12x12mm in its default configuration. So, you could scan an area about that size automatically, but you’d need to manually move the slide to get more than that. I started the [flat top stage] project to look at larger objects, but it currently doesn’t have automated XY translation. If someone was interested in designing an automated XY stage for the flat top version of the microscope (or indeed in adapting an existing XY stage design) that would be a really awesome instrument, I think.
Ah, 11x11mm should be doable - that’s about the limit for the OFM, but perhaps with a customised top plate to make sure it’s centred, that ought to work. I don’t typically do scans that large, so I’d be keen to hear how you get on.
That was useful to know. A 10x certainly works out a lot cheaper than 100x. I will order it right away, thanks!
We’ve been playing around the pollen sensing idea on and off since a couple of years.
Our goal is to build an autonomous pollen station that can collect and analyze pollen data in real time.
We’ve built a bunch of OFM recently and started just now to collect pollen images. Eventually we’d like to modify the OFM to integrate it in such ‘pollen sensor’.
Regarding the analysis of the images, I’ve a background in cell biology and I’m right now working on machine learning for classification of biological images, and I’ll try to translate what I know to pollen detection and classification.
It would be great to talk to you and start a collaboration!
Great! We are interested in all sorts of collaboration and a chat would be good. Has anything been published from your project that we could take a look at?
Our own first step will probably to put together a starter kit with non-printable parts and put it out on kickstarter.com. We were thinking of using a standard OFM with a 10x objective. But if you have a modified design maybe we should put that out instead?
This information about magnification, resolution and field of view can now be found in the handbook.
I have some very few info about the idea, but nothing online.
It’s really something that we worked on and off for a couple of years in the spare time.
The modified design was being still designed just before the lockdown, then everything was brought to a halt. However, we wanted to start easy by taking pictures with a plain OFM (also 10X objective) and testing the image analysis part. In the end we would like to make something like a Burkard pollen sampler, with everything integrated.
The kickstarter idea is great, but we’re really far from a finished product.
Where are you located? On wednesday at 20:30 CEST our community meets for a virtual call.
Hi! We also want to start simple. We are using the small device on the picture for collecting pollen. Using tape and small pc-fan. Then we use a regular OFM for taking pictures of pollen.Then finally we want classified the pollen using a some image recognition software. Obviously we are not very far either, but I think we can put out something on kickstarter to gauge the interest.
I am based in Norway.
Hey - cool to see that there are others playing around with this idea! We’re also doing pollen microscopy at Biotop in Heidelberg http://biotop-heidelberg.de
Last week we took these pollen (this time extracted from Honey) images with a 10x plan objective in the open flexure microscope and PiCam.
But professionally, my understanding is that a 40x-60x objective would be required, which we will try soon.
One issue we are facing is the travel of the microscope. We would need to scan the entire area of the 22mm square coverslip, but can only do about a quarter currently with the range of the stage. @r.w.bowman do you think the Delta Stage would have a wider range?
Also, does anyone have a suggestion of a reference database of pollen images to identify them?
The intended travel of the delta stage is similar, i.e. about 12mm laterally. However, making it taller should increase travel proportionally. 22mm might be pushing it, but 15 or 18mm might be possible - given that you can never really hit the edge of the coverslip (allowing for glue etc.) that might just be enough?