AMi: an Automated Microscope system for imaging plates


This site describes the construction and use of an inexpensive, automated microscope system for imaging samples in multi-well plates.  A paper describing the original system is here, and one describing a less expensive, more stable device that includes a dedicated graphical user interface is here. Updates to the design and software will be posted to this site as they become available. 

Please see the FAQs before writing me with questions.


We normally use this system to search for protein crystals in 96 well plates, but it can easily be programmed to image samples in 12 or 24 well plates as well. Plates with more than one sample at each row/column index are also supported. The computer-controlled stage yields reproducible 3D movements of 160 mm in x, 100 mm in y, and 40mm in z. To highlight what else is possible, here are a few images of some non-crystalline samples.




Hardware summary: (DETAILS)

The 3D translation stage is built from the light-duty CNC machine (~$135) shown on the right. The original machine allowed computer-controlled translation of the rotary cutting tool in X and Z, and translation of the object being cut along Y.

To convert the CNC machine to a 3D microscope stage, you’ll need to rearrange the parts, add some new, 3D printed parts, and screw the whole thing to stable base. You will also need to add a sample holder in place of the cutting tool, securely mount a microscope and camera, and connect the camera such that images can be triggered by the arduino-based board that drives the CNC machine. Construction requires a small amount of soldering, and some of the parts that hold the microscope and sample holder are 3D printed. You will also need access to a table saw and drill press.

The DETAILS page provides much more information.

white



Model 1610 CNC machine

white

Plans for the 3D printed parts are HERE.


As shown on the right, it is possible to use the AMi translation stage with an existing microscope. To fully integrate the system, you’ll need to mount the picamera (an inexpensive board-based sensor chip) in place of the camera shown. 3D printing files for a C-mount adapter that will couple many microscopes to the picamera can be found with the other plans below.

Be careful. Some microscopes may tip in this configuration! Also, any movement between the microscope and the stage will necessitate recalibration. The calibration procedure is described in the details link below If you’re using existing optics, my recommendation is that you mount the microscope pole to the base of the translation stage rather than have is physically separated as in the photo.





Current version of the AMi microscope system.

white text

Click here to see a movie describing the alignment procedure and showing the machine in action.



Connections between the component parts



Whit

e text




text


Software summary: (DETAILS)

3D printers and CNC machines use a text-based language known as G-code to control their movements. In the AMi system, a python-Tk based graphical user interface named AMiGUI controls the movements of the plate beneath the microscope and triggers acquisition of images of each sample. AMiGUI controls the translation stage through G-code commands. The AMiGUI software can be used to image an entire plate automatically, or it can be used to look at the sample drops manually in real time.

(HERE) is the link for the AMiGUI.py program. To run it type “python3 AMiGUI.py”

If you would like to take multiple images spaced in Z, you can do so by specifying the number of images and the spacing between them in mm. These parameters are entered in the lower right part of the graphical user interface. When multiple z-stacked images are collected, a command file for automatically converting these images to a singe image with digitally enhanced depth of field is written in the same directory. The script requires that the program Enfuse, which is part of the Open Source Hugin package be installed.

AMi can also image plates with more than one sample well at each row/column index (i.e. a 96 well plate with three sample wells at each of the 96 positions). See the DETAILS link for more information.

Using the system: (DETAILS)

Sample movement: The top part of the interface allows the plate to be moved beneath the microscope. Movements in X and Y are controlled by the part of the GUI that looks like a crosshair. Clicking in the Z-translation section changes the focus.

Alignment: The system needs to know the precise positions of the four corner samples before automated imaging can commence. The TL, TR, BL, and BL buttons move the view to the top-left, top-right, bottom-left, and bottom-right samples, respectively. At each position, users can make small adjustments in centering and focus and then click SET to save the save the precise coordinates.

Mannual operation: Once aligned, pushing the next and prev buttons move from drop to drop (i.e. A11> A12> B1> B2). Right-clicking these buttons moves up and down a row (i.e. D5 > E5). You can also enter a position (i.e. F11) and go right there. Left clicking the Snap Image button captures a single image. Right clicking the button collects a Z-stack of images.

Automated imaging: Clicking the RUN button initiates automated imaging of the entire plate. Normally a z-stack of images is collected at each position, and a custom command file is written for depth of field enhancement.




white

The AMiGUI graphical user interface is used to make manual movements, update the calibration file, manual inspect samples, take snapshots and automatically image the entire plate.





Image enhancement example:

The digital image enhancement serves two purposes:  First, it increases the depth of field.  Second, it relieves the necessity of having every drop perfectly in focus.  The software takes a z-stack (i.e. 3 images separated by 0.15 mm as shown) and creates a single image that contains the sharpest features of all of them.     The image on the right is one such image. The largest crystals are approximately 40 microns in length.

The three parent images are shown at below. These images are 2MP images, so the actual resolution is better than what you see.


Digitally enhanced image produced from three pictures separated by 0.15 mm.

white






   

   


    

Future plans:

I hope to make a variety of improvements to the device and I welcome input and technical help. 

Future plans are described here.


Disclaimer:

All warranties are disclaimed, including warranty of non-infringement.

Changes and updates:

The entire site was updated to describe the new version of the microscope with the GUI in July, 2019.