3D-printed microscope enables computational super-resolution imaging at $1200 | Hacker News Reader

3D-printed microscope enables computational super-resolution imaging at $1200

f1000researchdata.s3.amazonaws.com

85 pointsvo2maxer5 comments

Neat, hadn't seen this. I actually work in the research group at Uni of Bath that develops the original OpenFlexure design. If you have access to a 3d-printer and are interested in microscopy then I recommend building one. You don't have to spend $1200. There are different versions of the design you can choose from. Some just use the Raspberry Pi camera optics and are hand-actuated which makes them much cheaper.

https://openflexure.org/projects/microscope/

This is unfortunately just sensationalist at best, trying to milk "superresolution microscopy" with the other microscopy concepts, with this study using the 3D printed microscope as the gimmick.

To be clear, the OpenFlexure project is quite awesome and I am planning to build my own "garage" microscope sometime, and this design is the lead candidate. Two things you _cannot_ skimp on in a decent microscope, however, are a good objective and a good camera, and they both cost in the thousands at the least.

The constant struggle with biological microscopy (and I suppose most imaging technologies) is to try and extract the most amount of information possible for every photon your sample gives out. This will simultaneously allow you to observe samples without affecting them too much and get higher spatiotemporal resolution.

The whole field of superresolution started because we thought we hit the fundamental theoretical limits of how much resolution we can get from _any_ light microscope, even when using the cutting edge imaging technology. Superresolution techniques tried to trick and work around the limit by adding some extra imaging criteria (image one molecule at a time, screw with the illumination psf, fancy fluorescent proteins, etc). If you use suboptimal optics and then apply superresolution techniques, the image quality you get would just be worse with superresolution in this setup than with a regular microscope with good optics. That is actually clear in the figures in this paper - the "superresolution" images in the paper are worse than what any regular graduate student would get from a department confocal microscope.

Can't fault the particular study though: the entire superresolution microscopy field was a multibillion dollar blackhole of research funds with almost nothing to show by way of meaningful biological insight (leave alone deserving of Nobel prizes so soon).

Building on your comment, I think that while it does help to have high quality cameras and objective lenses, it is always worthwhile to push the limits of how cheap we can go. There have been many nicely done papers in this vein recently, showing for example, how cheapER cameras perform on supperresolution microscopy setups [1]. That paper showed that the setup is able to resolve cellular structures (actin filaments) less than 100 nm apart and measures the standard deviation of localization of single fluorophore emmitters.

Super-resolution implies that you are imaging below the diffraction (Abbe) limit. While the Fourier ring correlation can be used as a surrogate for resolution (it is frequently used in both cryo-em and fluorescent microscopy) you must define the correlation cutoff or else the measurement is meaningless. I do not see any mention of a cutoff used in the linked paper.

I would be more convinced if they actually imaged something smaller than the diffraction limit of their microscope, like the nuclear pore, which was recently proposed as a standard for the field [2]. Nonetheless I am impressed by the image quality given how un-rigid and prone to thermal drift/expansion their microscope probably is.

[1] https://www.nature.com/articles/s41598-018-19981-z

[2] https://www.biorxiv.org/content/10.1101/582668v1

Here's a great video presentation on the project:

https://www.youtube.com/watch?v=0F1pBmWuU3M

I'm very impressed with the hardware and software design of this project.

Beautiful, frugual sciences and accessible technics production (and affordable) is key to opening up biosciences to more people and possibly liberating costs to equipment making parts of biotech akin to what has become for computing when it comes to areas such as microcontrollers and others things (Thinking rapsberry pi and arduinos)

Reply on news.ycombinator.com