automated fluorescent microscope

[Michael Shamberger]

Hi,

I am interested in building an automated microscope that would be useful for synthetic biology.  Features would include automated tracking and classification of individual live bacteria.  Videos of the time lapse images would be transmitted and assembled in a PC and post processed with machine learning techniques.  I am a computer engineer so the robotics and machine learning are in my domain.  I want to understand the design for a DIY microscope in the sub 1000 range that would have the capability of creating the images.  

From what I have been researching on the web, I was thinking to build a LED Fluorescent microscope.  I am starting from the microscope design at openlabtools.org.  It uses a Rasberry pi and 5MB raspberry pi camera with an infinity optical system.  I want to add a fluorescent stage to that.  There are some DIY articles about LED Fluorescent microscopes but are lacking details though I continue to search.  If someone could point me in the right direction I would appreciate it.  I need information about where to source low cost LED's and design. 

Mike

[David Ng]

I have built a few LED-base fluorescent microscopes for custom research projects. One issue you will have to deal with is that the spectral bandwidth of a LED is very broad, i.e. although a blue LED might peak at 480nm, significant light will be emitted even at 510nm where you want to measure say, GFP.

That means you will definitely need dichroic bandpass filters on both the optical emission and excitation pathways, otherwise the unfiltered light will overwhelm the signal generated from the fluorescent specimen. As the light needs to enter the bandpass filter as parallel as possible, you will need to use a collimating lens on the LED in front of the bandpass filter. Its not all the band though, the LED and lenses can be bought for about $10, and you can actually get good bandpass filters on ebay for around $20.

-Dave

[Michael Shamberger]

Hi,

Thank for the reply.  I managed to find the Zephyr design put together by a IGEM team.  That is the high level design I am interested in so that I can have the microscope moving above the sample.

The issue there is definitely the cost issue of the filters costng $235, $235, and 190.  Can you give an example of a vendor on ebay that can supply these type of filters?  

http://2013.igem.org/Team:TU-Delft/Zephyr_How

520nm Bandpass Filter, 36nm Bandpass, OD6 Blocking, 25mm Dia, Stock No. #67-030	Emission filter	1	link 1
472nm Bandpass Filter, 30nm Bandpass, OD6 Blocking, 25mm Dia, Stock No. #67-027	Excitation filter	1	link 2
495nm Dichroic Filter, 25.2 x 35.6mm, Stock No. #67-079	Dichroic mirror	1	link 3

[Nathan McCorkle]

I'm working on this too, found the openlabtools data a few months ago
and ordered what should be an appropriate tube lens for monochromatic
use. Already have some infinity objectives from my current fluoro
'scope.

Personally I don't think the RasPi camera is too great for image
quality, but I'll be using it for it's high frame-rate for doing live
object tracking, position tracking, maybe other stuff. For getting
nice images to share with the world, best to find a different camera.
The digital noise on the RasPi camera is very appreciable, most
apparently under low-light conditions.

Why do you want to start with fluorescense? Is there anything you
actually want to see using this? It might be more interesting, if you
don't have a project that specifically requires fluorescense, is
building a monochromator light source. That way you can get full-color
images (of non-moving objects) but also get around chromatic
aberration. Or you can just get monochrome images of moving stuff. The
point is, getting around chromatic aberration is going to cost a few
hundred in lenses, I think.

> --
> -- You received this message because you are subscribed to the Google Groups
> DIYbio group. To post to this group, send email to diy...@googlegroups.com.
> To unsubscribe from this group, send email to
> diybio+un...@googlegroups.com. For more options, visit this group at
> https://groups.google.com/d/forum/diybio?hl=en
> Learn more at www.diybio.org
> ---
> You received this message because you are subscribed to the Google Groups
> "DIYbio" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to diybio+un...@googlegroups.com.
> To post to this group, send email to diy...@googlegroups.com.
> Visit this group at http://groups.google.com/group/diybio.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/diybio/6f496722-53f7-40c6-a754-b87e38643818%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.



--
-Nathan

[Michael Shamberger]

They are updating the microscope design this summer.  So far they have tracked a nematode.  

>> found the openlabtools data a few months ago 
and ordered what should be an appropriate tube lens for monochromatic 
use.

What did you end up buying?

>> Why do you want to start with fluorescense? Is there anything you 

actually want to see using this?

I want to see if I can track/image live bacteria with a DIY setup costing less the 1K. Then I want to try various machine learning techniques on the images to see if I can reproduce some of the latest research in this area.

[David Ng]

Well, with fluorescent imaging, you barely need machine learning as nothing else in the image is visible besides the labelled organisms. 

ie. for identifying normal bacterial colonies, I have to use a dozen or so filters, to correct for background illumination, do complex adaptive thresholding etc etc. For identifying fluorescent bacterial colonies, I just threshold all pixels that are 3x the standard deviation of the image. For bacteria, you could then just do a watershed, and filter on the size of the detected blobs.

As far as finding filters goes, search for "dichroic mirror" on ebay, and select the filter you need from the spectral info. for fluorescent imaging, make sure the excitation and emission bandpass ranges are seperated by at least 15nm to limit bleedthrough. If you are not doing multispectral imaging (eg GFP, YFP and RFP together), then excite at the blue end of the emission spectra, and use a longpass filter at the emission maxima of the fluorescent protein you are imaging. You'll need a dichroic mirror to separate the excitation and emission light too. It's cutoff should be in between the bandpass ranges.

You can look at the spectra and plan using this web tool:

https://www.lifetechnologies.com/de/de/home/life-science/cell-analysis/labeling-chemistry/fluorescence-spectraviewer.html

-Dave

[Will Sutton]

I'm also working on automating a microscope at bosslab!

Also using a Rpi, with stepper motors to turn the knobs. But I'd like to know of a less hack-y way to build robotics if I get to a v2.

And for camera, I'm using a Droid Razr as I find the camera autofocuses well at the very close proximity to eyepiece. 

For image processing, I'd recommend going with scikit-image package in python, here's some of the algo work to date:

http://nbviewer.ipython.org/github/sutt/BossLab/blob/willbranch2/science/summary1.ipynb

I think there's a big opportunity for DIY-er's to collaborate with code /github / ipython notebooks, even though we often can't meet.

~Will

[John Griessen]

On 07/08/2014 11:06 AM, Will Sutton wrote:
> I'd like to know of a less hack-y way to build robotics if I get to a v2.

Since you like python, there's micro python as a platform to consider, and if
you can see using a linux running board like RPi, there is ROS the Robot Operating System,
that can communicate with lesser microcontrollers, run motion and image recognition code.

Another motor controlling platform is smoothieboard, which uses ST micros much like
the micro python platform.

[Will Sutton]

Smoothieboard looks just like what this project needs, thanx John!

[Will Sutton]

Good resource: http://data.plantsci.cam.ac.uk/Haseloff/imaging/cheaposcope/index.html

On Friday, June 27, 2014 6:30:17 AM UTC-4, Michael Shamberger wrote:

[Will Sutton]

Team Aachen from iGEM 2014 used $2 emission filter to good results: http://2014.igem.org/Team:Aachen/Notebook/Engineering/ODF (see: "filter leaflet")

[Nathan McCorkle]

On Thu, Nov 13, 2014 at 12:18 AM, Will Sutton <wsut...@gmail.com> wrote:

Team Aachen from iGEM 2014 used $2 emission filter to good results: http://2014.igem.org/Team:Aachen/Notebook/Engineering/ODF (see: "filter leaflet")

Cool! Turns out the $2 item is a sample pack, and currently has a discount code for $0.01 (you still have to pay ~$5.80 for shipping). I just ordered two sample packs (though only one was $0.01) for less than $8. Looks like the company is targeting the movie/photo industry, rather than science... but they do offer a few dichroic filters, but didn't sell them directly ('contact a distributor').

[scoc...@gmail.com]

Not to go off topic but its incredible how the od600 led spec I was working on (and have a working prototype of) is identical to the item team's spec. Even the same led and initial design. So happy they did it with such good attention to detail and thorough understanding of optical density calculations. Very impressive work. :)

I did choose to use the light to freq converter instead of a photodiode and am designing a pcb using surface mount leds and an atTiny chip with a really good clock to count freq with high precision. Datasheets show that the photodiodes don't handle well at elevated temp. My idea was to make an ODlogging flask that you can throw into a 37c incubator and it will alert u via Bluetooth or wifi that your cells are ready. Either way super excited that "they beat me to it" and are using the device in their projects. +1 for small meaningful science tool dev!

Sebastian S. Cocioba
CEO & Founder
New York Botanics, LLC
Plant Biotech R&D

---

From: Nathan McCorkle
Sent: ‎11/‎13/‎2014 6:17 PM
To: diybio
Subject: Re: [DIYbio] Re: automated fluorescent microscope

On Thu, Nov 13, 2014 at 12:18 AM, Will Sutton <wsut...@gmail.com> wrote:

Team Aachen from iGEM 2014 used $2 emission filter to good results: http://2014.igem.org/Team:Aachen/Notebook/Engineering/ODF (see: "filter leaflet")

Cool! Turns out the $2 item is a sample pack, and currently has a discount code for $0.01 (you still have to pay ~$5.80 for shipping). I just ordered two sample packs (though only one was $0.01) for less than $8. Looks like the company is targeting the movie/photo industry, rather than science... but they do offer a few dichroic filters, but didn't sell them directly ('contact a distributor').

--
-- You received this message because you are subscribed to the Google Groups DIYbio group. To post to this group, send email to diy...@googlegroups.com. To unsubscribe from this group, send email to diybio+un...@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/diybio?hl=en
Learn more at www.diybio.org
---
You received this message because you are subscribed to the Google Groups "DIYbio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to diybio+un...@googlegroups.com.
To post to this group, send email to diy...@googlegroups.com.
Visit this group at http://groups.google.com/group/diybio.

To view this discussion on the web visit https://groups.google.com/d/msgid/diybio/CA%2B82U9Lk13Y8nXzLPK25VQhWGF2fs_mSMgPh_9xLg6d-2ZwRrA%40mail.gmail.com.

[BioBot]

hi Guys,

you should consider using Opentrons as a frame that has all the XYZ and it use the smoothie as controller with RsPI build it.

Since it's open source - you can just use the design .

Best,

C