Within two minutes of talking to Dr Richard Bowman, in his lab at the University of Bath, he’s guiding me through the physics of tractor beams in Star trek. He’s using it as a simile to explain the complicated subject of optical tweezers to a stupid person.
He does so in a charming way, as someone familiar with explaining his complex field to journalists, but it’s clear why he’s a Prize Fellow and Royal Commission 1851 Research Fellow – his explanation ends with our imaginary tractor beam melting an object it’s trying to move before Bowman shrinks this entire sci-fi example down to demonstrate how he’s used laser beams in his past work to move tiny objects.
On the face of it, Dr Bowman’s story is that he’s on a three-year project to build a general-purpose 3D-printed microscope, but his ambitions are bigger and, ultimately, he wants to create 3D-printable ‘building blocks’ that others can use to make affordable, new experimental apparatus. “Opening up hardware means more people have access to it,” says Dr Bowman.
“Personally, I’m keen on pushing the hardware to be of a quality that I would not be embarrassed to use in a well-funded university research lab. There’s also a big push to having stuff you can make at home, and so making scientific research more accessible means that schools, science clubs and makerspaces can start doing really interesting scientific stuff. […] The more of the public that gets involved in science the better, as far as I’m concerned.”
As well as being a co-founder of WaterScope, a project developing faster and easier-to-use water-testing field kits that use a microscope he designed, he’s a passionate advocate for open hardware and works with organisations like GOSH (the Gathering for Open Science Hardware).
The journey begins
Dr Richard Bowman’s journey towards open hardware began while working on optical tweezers for his PhD at the University of Glasgow. There he experienced first-hand the cost of custom scientific instruments. A research-grade microscope “with all the bells and whistles and a motorised stage”, says Bowman, will set you back £30,000 to £40,000. “Then you void the warranty by ripping out most of the complicated optics from inside of the microscope and replace it with your own stuff.”
The situation was frustrating and not as efficient as it could be but then – at Queens College, Cambridge University, where he was working in a nanophotonics group and dealing with automated microscopy – he began meeting people interested in open source hardware: “I met someone who was building a 3D-printed microscope and it looked a lot like this,” says Bowman pointing to a RepRap 3D printer nearby.
That was Alexandre Kabla and a project called OpenLabTools. “The goal is to be self-replicating and to print as many of its parts as is possible, but in practice you found that most of it isn’t printed,” says Bowman, and that was the spark that got him thinking: “I was curious just how much of a microscope you could print.”
Bowman shows us one of the first microscopes he printed: “This,” he says, “will turn a Raspberry Pi camera into a microscope.” It’s a tiny black extension tube (pictured in the image at the top of the page), which photographers have used for a long time: “On your webcam you have a tiny little shiny silicon sensor and the pixels on there are very small […]. I think it’s 1.1 micron [across] for the version 2 Raspberry Pi Camera Module. So this lens that forms an image on the sensor is, in fact, a microscope objective, because it’s focusing the light down to a spot not much bigger than a micron and for linear optics at least, […] you can reverse the light path and it does the same thing.”
Dr Bowman explains that if he unscrewed the lens and pointed it at the object he wanted to look at – putting the lens some distance away from the sensor – it would function as a microscope.
Surprisingly, the difficult aspect of microscopy isn’t the optics but the mechanicals. Working with objects that are millionths of a metre across requires a very high-powered microscope to see anything and at that point the depth of focus of your microscope is less than a micron. Using a larger motorised model of his 3D-printed OpenFlexure microscope (pictured above), Bowman demonstrates the problem. “If your sample wobbles even by a micron […], a hundredth of the width of a human hair, your whole experiment is ruined.”
The expensive part for serious science work, then, is building a mechanical stage for fine control of what you want to look at. “A Raspberry Pi Camera Module is £25, but the mechanical stage might then cost you £1,000 or easily more than that.” So Bowman began a process of researching and prototyping mechanical stages. You can see some of the iterations pictured in the image at the top of this page: the key step up, which Bowman describes as a “middle of the night” epiphany, was having the sample sit on a table (the last red microscope in the top image) that has legs designed to bend in a way that allows for the crucial focus control and movement on the X and Y axes.
Ultimately, Bowman’s microscope design is a complicated structure and impossible to machine: “You can print it layer by layer, but you can’t machine it,” says Bowman, smiling. “You couldn’t injection-mould it either.”
Bowman’s larger microscope with a motorised stage, which he demoes to us, uses Python scripting for enabling the user to move the field of view around the slide. The aim is to add more features such as auto-focus and the ability to stitch all the images back together for a digital representation.
WaterScope, the company that Bowman co-founded, came out of a student project in the i-Teams Cambridge scheme (that includes the Raspberry Pi as past successful projects). Part of this scheme focuses on innovations that can be used in the developing world to improve people’s lives in a sustainable way.
Globally, 663 million people don’t have access to safe drinking water, so WaterScope decided to target testing for bacteria in water as it’s such a lengthy and fiddly process: “Testing for bacteria is difficult and inconvenient, so this means people don’t do it as much as they should,” says Dr Bowman. “You collect your water; you extract the bacteria from the water with a filter, you pass it through a filter with very small pores, less than micron. Next, you put your filter on some medium, so food, and you let them grow overnight and 18 hours later you look for the biggest, nasty bacteria splotches.
“The colonies will be big enough to see by eye […]. If you’re working in a developing country where the infrastructure isn’t good, you probably don’t have access to a clean lab and if you do it’s probably some distance from where you are water testing. So you end up gathering your samples, going back to base and then you’ve got to sterilise everything, keep your work area clean and so on.”
The idea behind WaterScope is to make the test faster by doing it on a smaller scale. After all, says Bowman, “if you look at it with a microscope, you can see single bacteria directly,” and the equipment is small and portable and doesn’t require a lab and hopefully can be a little bit more digital. Bowman says if you can grow the bacteria, it should be possible to tell what sort of bacteria you’ve got, and that last part is something WaterScope is actively working on now.
The idea is to acquire a time-lapse video of an hour’s growth and then process it so you can automatically count the number of bacteria colonies. “You can save enough of that image data to verify later that everything has worked properly,” says Bowman.
WaterScope’s eventual goal is for technicians to be able to upload and map out the tests and turn it into a useful resource for people planning water provision. “We’ve now done our first proper field trials this summer out in Tanzania,” reveals Bowman. “We have seen bacteria growing. We’re confident we’ve got something that works, but we’re still working on how to identify what bacteria you’ve got in the field. The trouble is that a lot of the existing systems assume you’re doing a really long incubation and they only work after 4, 8 or 10 hours – we really want something that’s much faster, but it’s exciting progress.”
Acceleration through automation
Another exciting thing Bowman is working on involves the higher-specified version of his microscope. He points out some blood smears: “Those blue things inside the red blood cells,” he tells us. “Those are malaria parasites. And this is how a lot of malaria is diagnosed,” he says. “They will take a drop of blood and dry it out with some stain and someone will spend half an hour looking through your blood cells for those tell-tale rings.”
Bowman is pushing hard to automate this process: “All my project proposals are working with computer vision folk to make this process more automatic […]. The big benefit [to doing this] is that there aren’t enough technicians […]. If this means that one technician can have ten microscopes running simultaneously because they are all automated then great. […] If they refer a patient on to a hospital then they can send the slide digitally or even on paper, so they don’t have to redo the microscopy.”
But the microscope is one part of the open science hardware dream: “One of the reasons I put so much effort into making the microscope print more or less in one piece on a bottom-of-the-range RepRap 3D printer is to get as close to that dream of being able to download and have a piece of hardware.” And Bowman admits it’s pretty close. “It takes maybe half an hour to put one of these together.”
Currently, Dr Bowman is collaborating on the computer vision with another academic at Bath University, Dr Neill D.F. Campbell, and they are currently applying for funding.
(This feature was first published in issue 185 of Linux User & Developer).
- Subscribe to Linux User and Developer today! Get an extra 20% off our Xmas subs deals using code: 20NOV. That’s up to 43% off PLUS an extra 20% off!
Powered by WPeMatico