Host: Shamini Bundell
Welcome back to the Nature Podcast. This week, super-quick computer switches…
Host: Nick Howe
And a new algorithmic way to measure heart health. I’m Nick Howe.
Host: Shamini Bundell
And I’m Shamini Bundell.
[Jingle]
Host: Nick Howe
So, listeners, as you may know, we are all working from home at the moment due to the coronavirus outbreak, so that might be why things sound slightly different. For instance, I’m currently in a pillow fort recording this. Shamini, where are you coming to me from?
Host: Shamini Bundell
Oh, mine’s one step up from that. It’s actually a sofa cushion fort. It was carefully constructed by, well, initially by me and then it fell down and my flatmate who’s an engineer had to come in and rescue me and explain how buildings work.
Host: Nick Howe
Laughs. How buildings work.
Host: Shamini Bundell
When they’re made of sofa cushions. She made it stable, is the important point. I now have a structurally stable sofa cushion studio so that I can record these lovely podcast links.
Host: Nick Howe
Well, as long as it’s structurally stable enough that you manage to get through the rest of the show with me, I think we’ll be good.
Host: Shamini Bundell
Hopefully.
Host: Nick Howe
So, with the obvious audio differences in mind, we’re keeping the main Nature Podcast as normal as humanly possible, and for a little light relief, we’re making it somewhat of a coronavirus-free zone. If you are interested in coronavirus updates though, then please check out our new show, Coronapod, coming to you on Fridays. Back to the Nature Podcast now though, and Shamini, what’s coming up first this week?
Host: Shamini Bundell
Well, first up, we are talking about switches. So, switches are pretty useful in electronics, and not just for starting up your games console. The very zeros and ones that make up the digital world are created by switches known as transistors, which turn tiny voltages on and off. But not all switches are equal. In some transistors, going from off to on might take a billionth of a second, and while for many applications that’s quite fast enough, for others, switching on or off extremely quickly is crucial. Now physicists have found a way to use a surprisingly simple mechanism to make an ultra-fast electrical switch. It can flick on a sizeable voltage in just a trillionth of a second, and that can be used to generate radiation that has exciting uses in things like medical imaging. Nature reporter Lizzie Gibney spoke to physicist Elison Matioli to find out more. She started by asking him how switches are used in the world of electronics.
Interviewee: Elison Matioli
So, switches are widely used, most frequently in the form of transistors. They switch from a conducting state to an insulating state, so you can transmit information. You can also do calculations. You can do a lot of different operations using switches.
Interviewer: Lizzie Gibney
And in your work, you’ve come up with a switch that is incredibly fast. Why might you want a switch to be fast? What are the advantages of that?
Interviewee: Elison Matioli
So, switching fast is extremely important in many different applications, but in our case, we’re interested in generating extremely fast electromagnet radiations, and if you can switch extremely fast electric fields, you can produce a wave that propagates in a very high frequency. Our goal was to generate waves in the terahertz.
Interviewer: Lizzie Gibney
So, why can’t we use regular electronic switches to make these very fast transitions?
Interviewee: Elison Matioli
So, regular electronic switches, nowadays, are based on semiconductors, and the semiconductors have some limitations in terms of how fast electrons can flow inside them, and that limits the switch speed and the output power that they can transmit.
Interviewer: Lizzie Gibney
So, you want to create these very high-powered but very, very fast switches. What approach did you take?
Interviewee: Elison Matioli
So, our approach was completely different than the typical approach that exists today. So, instead of using a semiconductor element, we basically used just two pieces of metal spaced by a very, very narrow gap in between. So, the idea is that if you apply a small voltage, there’s no conduction between the two paths, so electrons don’t flow between them. And when you apply enough voltage, electrons start conducting through air or through any kind of gas that you have in between.
Interviewer: Lizzie Gibney
And how did the electrons actually flow then? So, you have a field so strong that suddenly, they literally are wrenched out of one and go into the other.
Interviewee: Elison Matioli
Exactly. You apply just enough electric field so you can extract the electrons from one metal and they come to the other one.
Interviewer: Lizzie Gibney
And can you give a sense of just how fast the switch works, how quickly this change from insulating to conducting happens?
Interviewee: Elison Matioli
The switches are extremely fast. So, we could, in our laboratory, measure up to 12 volts per picosecond. That’s 12 volts in a trillionth of a second. This is more than 10 times faster than any semiconductor device that exists today, and this has more than 200 times higher power than what’s possible with a semiconductor device today. But that’s what we could measure, because, to be honest, we’re limited by the speed of our measurement setup. So, the same way that making an electronic device that it can produce the signals, it’s extremely challenging. We don’t have any device that can measure these extremely high speed signals, so we oftentimes need to do indirect measurements to know where we are.
Interviewer: Lizzie Gibney
So, you’ve got this switch that happens very, very fast. What then are you going to use that for?
Interviewee: Elison Matioli
So, these devices are extremely simple. They consist just of two pieces of metal very close together, and they can be easily integrated into many kinds of devices. They can be integrated with antennas if you want to radiate these signals. So, once you have these high frequency waves being radiated, you can do many things. Of course, communications is an obvious one because the higher the frequency, the more data you can transmit. But there are many other applications where terahertz waves are extremely interesting. For example, in imaging. These kinds of waves are non-destructive waves. They are non-ionising. So, when they go through biological materials, they basically don’t do any damage to the tissue or cells, so it’s a nice way to do bioimaging without causes any damage to biological materials.
Interviewer: Lizzie Gibney
Are there any applications that aren’t involving using this very fast switch to create a high frequency wave? Could you use just the switch in itself for anything?
Interviewee: Elison Matioli
There are many applications that we can use the switch in itself. For example, one application that we are looking at is in protecting devices. So, those switches, they can turn on and off extremely fast, so if a high voltage is applied to a very sensitive device, they are much faster than the device itself.
Interviewer: Lizzie Gibney
And could we use a device like this in a computer like we use a transistor today?
Interviewee: Elison Matioli
So, these devices are not meant to replace transistors. So, transistors, they have an extra terminal to control their conductive or insulating states. These devices are a lot more simple, but it’s meant to be a different way to generate high frequency and high power waves.
Host: Shamini Bundell
That was Elison Matioli of the Swiss Federal Institute of Technology in Switzerland talking to Lizzie Gibney. Elison’s paper is online now on the Nature website, and we’ll provide a link in the show notes.
Host: Nick Howe
Next up, it’s time for the Research Highlights, coming to you direct this week from Dan Fox’s pillow fort.
[Jingle]
Dan Fox
Listen to this.
[Sound of ice breaking]
That was the sound of a huge chunk of ice breaking off of a glacier in Svalbard, Norway, recorded from underwater. Now, exactly how big do you think it sounded? That was the question the researchers who recorded this wanted to find out. They used underwater microphones to record the sound made by ice breaking off of the glacier and into the ocean. They could then use that audio to extrapolate the amount of ice that broke off each time. They think that underwater microphones could provide a useful method for tracking how much ice is being lost in events which are often too small and frequent to see on satellite images. That’s just the tip of the iceberg. Read the rest of the paper at Cryosphere.
[Jingle]
Dan Fox
In the next ten years, automated flying drones will be delivering everything from insulin prescriptions to chicken jalfrezi, if you believe some news coverage, at least. But this drone-driven future could be a step closer to reality, thanks to a team of researchers at the University of Zurich. They’ve developed a flying drone capable of dodging obstacles ten times faster than existing ones. To test its dodging ability, researchers did the obvious thing – pitching balls at it. They found that it started to dodge an incoming ball in just 3.5 milliseconds, significantly faster than conventional drones. To accomplish this deft dodging, the drone is fitted with a type of motion detecting camera called an event camera. Instead of recording images, an event camera outputs a stream of data points triggered by changes in environmental brightness, allowing the drone to identify moving objects much more quickly than if it was using a regular camera. Don’t dodge that research over at Science Robotics.
[Jingle]
Host: Nick Howe
Next up on the show, is there a better way to measure cardiac health? Ali Jennings has been getting to the heart of the matter, and he comes to us now from under a duvet.
[Beating heart sounds]
Interviewer: Ali Jennings
That’s the soothing ‘lub-dub’ of a healthy heart. The ‘lub’ sound comes when the ventricles of the heart, the chambers that pump blood out into the body are at their fullest. The ‘dub’ comes after the ventricles have contracted at their emptiest. If you compare the volume of blood in the ventricles before contraction to the volume after the blood has been ejected, you get a useful measure of the heart’s health – the ejection fraction. The high ejection fraction shows that the heart is pumping strongly. A low fraction suggests that the heart muscle could be weak. In this way, the measure provides an early warning sign of heart problems. To calculate a person’s ejection fraction, the most common method is to take an ultrasound of the heart. This produces a grainy, black and white video of the patient’s beating heart. The clinician then goes through the video, searching for the moment when the left ventricle is at its largest and when it’s at its smallest. They must then outline the two areas by hand for comparison.
Interviewee: James Zou
It’s actually pretty labour intensive to trace out the heart and to find the right frames.
Interviewer: Ali Jennings
This is James Zou, a computer scientist from Stanford University in the US.
Interviewee: James Zou
And that’s why typically a clinician would only do this for one beat of the heart, where actually it’s recommended that one should actually do this across multiple beats to get a better sense of the average cardiac function.
Interviewer: Ali Jennings
To speed up this laborious process, this week in Nature, James and his team demonstrate an algorithm that can analyse echocardiogram videos and calculate the ejection fraction all by itself. And the algorithm seems to have an edge over the clinician’s.
Interviewee: James Zou
It’s actually very easy for it to look at all of the beats from an ultrasound and compute some sort of average ejection fraction across all of the beats, and that’s actually really clinically important because for some heart issues, there’s sort of variability across different beats in the same patient, and if we only trace that one beat then that could actually be missing a lot of that important variability.
Interviewer: Ali Jennings
The algorithm does the exact same thing that the clinicians do, but it can do it over every beat in a video in the same time it takes a clinician to draw out one beat. Then it can calculate an ejection fraction that is more reliable and accurate. This means better diagnosis and treatment for people with vulnerable hearts. The next step for James is to make the algorithm adaptable, able to work in any hospital and with any equipment.
Interviewee: James Zou
We can also show and can train it to also be robust across different hospitals and also robust across different ultrasound machines. Of course, I think there’s still a lot more work that we need to do, and we are actually working on that to really make sure this generalises across other countries, across very different environments, especially in low-resource settings.
Interviewer: Ali Jennings
James hopes his AI will be especially useful in places like this, which might lack trained cardiologists. And eventually, James imagines his AI having even broader applications.
Interviewee: James Zou
I think we will love to, first of all, have this tool, extend it so that based on the ultrasound videos we can predict other diseases, for example, diabetes, and other heart failures. So, that’s something that we hope to be able to train the algorithm to be able to predict going forward.
Interviewer: Ali Jennings
For now, though, James’ team will start running pilot tests on the Stanford population this year, and hope to start testing in other hospitals and countries in 2021. So maybe, in the not-too-distant future, you’ll find your cardiologist has teamed up with their computer to help keep your heart healthy beat by beat.
Host: Nick Howe
That was Ali Jennings talked to James Zou. You can find James’ paper at nature.com, and we’ll put a link in the show notes.
Host: Shamini Bundell
Normally, right now, we’d have the News Chat, but as I’m sure you can appreciate, everything is rather coronavirus-focused at the moment, so for more on that, you can check out our new Coronapod, coming this Friday. And that’s pretty much it for this week’s show. Do get in touch with us if you want to let us know how you’re dealing with the outbreak or if you have any fun distractions that we can share.
Host: Nick Howe
And to do that you can reach us on Twitter. We’re @NaturePodcast. Also, you can get in touch with us by email at podcast@nature.com. I’m Nick Howe.
Host: Shamini Bundell
And I’m Shamini Bundell. See you next time.