MicrobeHunter.com Microscopy Forum

I am pretty sure, it is way too expensive but is still interesting... Has anyone used it, any opinions?
https://www.aurox.co.uk/clarity.php

Thanks for posting. Interesting, even if not really affordable.

They claim somewhere that the device fits modern and classical (if I am not wrong) microscope - I wonder how back in time that means... do they specify the mandatory features of the microscope ?

As far as I am understanding, there is no limits but infinity optics and even that may not be strict limit.

At work we've demonstrated similar systems (high-speed structured illumination), although not this exact model. They work via the SIM principal to provide confocal (or slightly better) resolution. These are probably outside of the home users budget though - the modules are generally 10's of thousands of dollars, and require a scope with infinity/high NA (>1.40) objectives. They also typically use EM-CCD or back-thinned CMOS cameras, which again tend to be tens of thousands of dollars.

Cool toy of you can afford one though!

Brings up the question, is that all still too complex for amateur level open source hardware approach - light modulators are reasonably cheap now, the same has pretty much happened to back thinned sensors.

Something like this would likely be beyond an amature. But there are a lot of systems being designed that you can make at home with 3D printing (UC2 and flexscope being two example) that can be configured to provide very high flourescence resolution - upto 20 nm using dSTORM, and even HALO illumination.

The tech note, how the Aurox system works:
https://www.aurox.co.uk/downloads/broch ... oscopy.pdf

If I understand correctly, the processing needed is literally just image differential, thus low computational requirements and fully realtime. It would need good software alignment of the images but the idea itself seems quite brilliant in the simplicity - 50% reflection striped rotation disc is pretty much the only critical component.

The original publications mentioned in the tech notes:
Jus̆kaitis, R., Wilson, T., Neil, M. et al. Efficient real-time confocal microscopy with white light sources. Nature 383, 804–806 (1996). https://doi.org/10.1038/383804a0

Nice interview with one of the system authors, Tony Wilson:
https://analyticalscience.wiley.com/do/ ... 1918/full/

Seems a bit over the top. I think this may cost more than the confocal microscope it is designed to emulate.

It seems not to be that complex and has benefits over the classic confocal systems, especially for amateur use with cheaper cameras. There is Zeiss version also, called VivaTome:
https://zeiss-campus.magnet.fsu.edu/art ... ction.html

Greg Howald wrote: ↑

Sun May 22, 2022 11:20 am

Seems a bit over the top. I think this may cost more than the confocal microscope it is designed to emulate.

The SIM light pattern can be projected from DLP projector, just changing out the light source to specific narrow spectral range LED:

https://www.nature.com/articles/srep01116

As far I understand, this gives the same properties for high resolution sectioning as the spinning disk pattern. Higher than 1.4 NA objectives are not required, although would be interesting for the highest resolution possible.

Complexity seems to be minimal, basically it is just modulated light source with post processing for sectioning image recovery:

Really nice video on SIM principles, by David Agard:
https://www.youtube.com/watch?v=i73HhpLJrqs

Basically it is boiling down in information gathering on wavefront, increasing NA with tricks beyond what is possible with just glass shaping in objective. Appears as simple as changing the light source with DLP projector module and a little of postprocessing. Texas Instruments has multiple development kits, that are starting from 100 USD mark; better ones around 1000...2000 USD have internal 1D or 2D pattern control abilities. There seems to be MicroManager support for DLP devices. Anyone here already using DIY DLP based structured illumination or can refer interesting DIY project link?

Up to 1920x1080 resolution, trigger input and output, 1D pattern (seems to need blanking frames, optics offset may be a problem):
https://www.ti.com/tool/DLP4710EVM-LC

Up to 1920x1080 resolution, all bells and whistles up to real continuous static pattern (but more expensive):
https://dlinnovations.com/products/dlp-discovery-d4100/

Github project for lightmodulator modification:
https://github.com/WitsOCLab/DLP4710-SLM

What considers suitable fluorescence cameras - cheap cooled astronomy cameras seem to be beating 5 year old 12k professional sCMOS camera on readout noise and sensitivity. Sure, relatively low frame rate but this is more the question of sensor resolution.

IMX533 camera - cooled, mono and quite remarkable:
https://astronomy-imaging-camera.com/pr ... i533mm-pro

3008*3008 pixel
3.76um pixel
Diagonal: 15.97mm
QE peak: 91%
Full well: 50000e
Readout Noise: 1.0 to 3.4e-
Dark Current: -20C，0.0005e- /pixel/sec
Exposure Time Range: 30us-3600sec
Native 14-bit A/D
3008×3008 19.88fps
1920×1080 54.11fps
1280×720 79.74fps

1159 usd - not cheap, sure. Not end all expensive either.

About the SIM image stack processing - I have yet to try it but there is free to use machine learning code up on Gitlab and a nice 2021 publication going with it:

ML-SIM
https://github.com/charlesnchr/ML-SIM

ML-SIM: universal reconstruction of structured illumination microscopy images using transfer learning
https://opg.optica.org/boe/fulltext.cfm ... &id=450173

From the publication: "The repository also holds source code for a desktop program with pre-built installers for Windows, macOS and Linux. The program makes it easy to use ML-SIM and perform batch processing via a graphical user interface. Includes a plugin for μ Manager [42] that enables a real-time live-view of ML-SIM reconstructed output during acquisition"

I would say, that it is all literally delivered on silver plate - complicated components are there, ready made (DLP projector, camera, software with microManager plugin) - only user task is to have the structured light illumination coupled to the microscope (this may be easier said than done though).

As expected, someone has already realized quite nice example of optical setup with the DLP4710 EVM:

http://isopack.blogspot.com/2022/03/ima ... scope.html

About the devils and details:

Texas Instruments DLP evaluation kits have image offset 100% - for symmetric projection to field diaphragm plane one needs to reposition projection lens and preferably scale the projection image size as necessary. Photographic tilt shift lenses are useful ready made solution there; for projected image scaling reversed prime lenses with different focal length than tilt shift lens work.

Really nice repo:

"... contains code for designing, analyzing, and carrying out multicolor structured illumination microscopy experiments based on a digital micromirror device (DMD-SIM), including DMD simulation code, DMD pattern generation, SIM reconstruction and instrument control. It also includes a number of useful utilities for simulating the resulting diffraction pattern given certain DMD patterns, determining system point-spread functions and optical transfer functions, and determining the affine transformation between the DMD coordinates and the imaging space coordinates."

Texas Instruments DLP6500 direct control code etc, blaze/diffraction angle optimizer for multi-color etc.

https://github.com/QI2lab/mcSIM