The World's Tiniest Camera
Soon medical cameras will be as small as grains of sand--and almost as cheap
Steven Cherry: Hi, this is Steven Cherry for IEEE Spectrum’s “This Week in Technology.”
Has anything changed as much and as quickly as camera technology? Cameras are like the ice that comes when you order a Coke, an expected feature for which you pay basically nothing. They’ve gotten incredibly small, and incredibly cheap. What they haven’t been, until now, is incredibly small and incredibly cheap. That’s about to change.
The world’s smallest digital camera is now built directly on a silicon chip. Even including its tiny glass lens, it measures just 1 millimeter by 1 millimeter by 1½ millimeters. That’s about the size of a pinhead, or an extralarge grain of salt. Sea salt, I guess. Eventually, I guess, pinhead cameras will be embedded in our eyeglasses, our clothing, maybe everywhere. But the first ones will be used in endoscopic devices—those snakelike cables doctors use to send a camera into a colon or a lung without cutting a patient open.
Martin Wäny is the CEO of Awaiba, an imaging technology company in Funchal, Portugal, and he’s the project manager for the development of these supersmall camera sensors, which Awaiba is making in collaboration with researchers at the Fraunhofer Institute for Reliability and Microintegration, in Germany. He joins us by phone from Portugal. Martin, welcome to the podcast.
Martin Wäny: Oh, thank you for having me, and I’m pleased to talk with you.
Steven Cherry: Good. Martin, I’m not sure our listeners can even picture this camera. I saw one picture that had it next to a wooden match, and the match was huge compared to it—it was as if one was a car and the other was a motorcycle. Maybe just start by describing the camera itself.
Martin Wäny: Well, the main feature of the camera as you said in your introduction is that it is small. I remembered when we received the first prototypes of the chips, they came in a gel pack from dicing. It was a little bit special kind of dicing because it’s so small. Eventually, I opened the gel pack here in the lab, and I got very upset because I could not see any chips in there, and I thought, “How come have they forgotten to put anything in that gel pack?” Only at the second look I could see that there were the dies in between the grid of the gel—we could see them. So if you handle a camera and you drop it on your carpet floor, you have a really hard time to find it. Other than that, it’s a digital camera; the main application is medical endoscopy, so the lens we have is a relatively wide opening angle. It is about 90 degrees opening angle, so it’s quite a wide-angle device, which did not make it easier but it permits the doctors to have a good overview. And the camera is digital and provides the data over a two-wire LVDS [low-voltage differential signaling] interface and has two other wires for electrical power supply.
Steven Cherry: Yeah. I guess the data wires themselves are incredibly thin, and I guess you had some trouble attaching them to the camera. Is that right?
Martin Wäny: Indeed, it was one of the more difficult tasks, was to find a good process to attach the cabling to the camera without making the overall camera much larger, especially the diameter. If you think you would reflow solder these on a PCB board—the smallest PCB board you could cut out and having some overhead on the sides—you would need to have four or five times the cross section of the camera by itself. So we needed to find a rather special process of attaching the wires to the camera.
Steven Cherry: Let’s just talk about the camera itself. What was the breakthrough that made it possible to make these relatively cheaply?
Martin Wäny: The silicon is produced on wafer level, but also the packaging is done on wafer level; the optics are produced on wafer level. What is different and what we are probably the first company in the world to do is that we actually assemble all these three steps, wafer by wafer, together and then have full wafers with the cameras on there. We have to dice them in parts, but then on one wafer we get a lot of these cameras.
Steven Cherry: So in effect you’re making cameras as if they were just chips.
Martin Wäny: Exactly. Basically, we use VLSI [very-large-scale integration] production technologies for making the complete cameras.
Steven Cherry: You mention the very wide angle—90 degrees. The data itself is 62 000 pixels. That’s a lot less than even a mediocre cellphone.
Martin Wäny: It’s about a 250- by 250-resolution. This is for the special application. It’s actually quite a high resolution; it’s not the megapixels you would be used to from your iPhone camera. But in endoscopy, it’s critical that you have real data in each pixel and that we have a relatively good light sensitivity, so we did not go for the smallest possible pixel size.
Steven Cherry: I noticed that the frame rate was pretty high—44 frames per second. That’s more than even a movie in a movie theater.
Martin Wäny: Exactly. And that’s for various reasons. One reason is if you use these devices in a surgical context, you have the object very close to the lens, so you get kind of a microscope effect—any motion will be magnified. So if the endoscope moves a little bit, the relative motion in the image will be higher than if you would have an [unintelligible] imaging that you would be doing, for example with your camcorder. So having a relatively short exposure time is vital to have a crisp and clear image. And then the other reason for having a relatively high frame rate is to—on a scope context—reduce the overall delay between capturing the image and displaying a digital image on a screen, because this is kind of a closed-feedback system. A doctor has to look at the image and then readjust the position of his instrument or the position of his scope to perform a surgical procedure. And if you have too much delay in that, it feels like not being very accurate, and it’s getting more difficult for the surgeon to do a precise operation. So this is a requirement quite typical for endoscopy to have a relatively high frame rate.
Steven Cherry: Now this isn’t a lot smaller than current endoscopes, is that right? It’s just that you’re making them so much less expensive.
Martin Wäny: Well, it’s probably a bit smaller than most classical optical endoscopes. In this category where these 1 millimeter devices go, currently you only have fiber scopes. However, the fiber scopes—they usually have an even much lower resolution—in the same diameter, a fiber bundle can have maybe 10 000 to 15 000 image points at max, where we have 60 000 image points in the current NanEye camera. So we have a quite significant improvement in spatial resolution. And on the other hand, the device is possibly very cheap, so that we can go for disposable endoscopes, while a fiber bundle of 10 000 to 15 000 fibers is already quite an investment, and you could not consider to use it just for one procedure. You have to sterilize it, you have to maintain it; if the doctor runs over it with the chair or the trolley, it is broken and it makes a whole disruption in the whole surgical plan. So the fact to have a digital small device which is more flexible and also more cheap is overall a quite significant improvement.
Steven Cherry: Very good. What are some other applications that we might expect for these supersmall cameras?
Martin Wäny: In medical devices still you can think of adding vision to instruments that previously did not have vision either because it was too big or too costly. Some examples of these would be intubation devices, guide wires that would go to some location, for example to the lung to place an RF identifier coil or to make a way to bring a balloon or another instrument to a certain place. Then outside of the medical area you can think of inspection in tiny spaces. You can maybe think of spy kits, undercover agents, or you could just almost integrate a camera anywhere, because space is not really a limitation anymore. You just have to make sure you can bring the data somewhere to a central point where you can do something with the data, but mechanically mounting it is not really an issue anymore.
Steven Cherry: Yeah, you mentioned intubation. I guess it would be possible—I saw on your website—to count individual drops of, say, a medication.
Martin Wäny: Yeah, that’s another application that we are working on that goes more to the classical machine vision. However, if you have an automatic dispenser, for example for DNA analysis, this is a massively parallel dispensing [unintelligible] and you could not have a microscopic camera to observe that each dispenser really dispenses a drop when it’s supposed to. So I think a very small camera to the head and also a relatively inexpensive camera to the head makes it much more practical to use a machine vision system to control that each position drop is dispensed.
Steven Cherry: Let me just ask you—you know, these are so incredibly small. Are you approaching some sort of absolute physical limitations on how small cameras can go?
Martin Wäny: The optical wavelength does not really scale down with Moore’s Law, so at some point it will not make sense to make optical lenses anymore if we get too small, because the resolution would just decrease. But if you accept that your image may only have a thumbnail resolution, I think it could be going much smaller still in terms of nanomedicine or nanorobots.
Steven Cherry: Very good. So the current camera—the millimeter camera—when will we expect to see those devices on the market?
Martin Wäny: We have first customers that are actually sampling products with the first generation of this camera to doctors. A lot more applications I would expect in a time frame of about one to two years to come to the market, just because regulatory approval is a bit slow in the medical device industry, so once we have a device out, it takes some time until it’s in the marketplace.
Steven Cherry: Very good. Well, it’s a tremendous achievement, Martin, and congratulations. And thanks for speaking with us.
Martin Wäny: Yeah, well, thanks for having me. It was a pleasure to talk to you.
Steven Cherry: We’ve been speaking with Awaiba CEO and imaging technologist Martin Wäny about engineering the world’s smallest digital camera.
For IEEE Spectrum’s “This Week in Technology,” I’m Steven Cherry.
This interview was recorded 25 April 2011.
Segment producer: Ariel Bleicher; audio engineer: Francesco Ferorelli
Follow us on Twitter @spectrumpodcast
NOTE: Transcripts are created for the convenience of our readers and listeners and may not perfectly match their associated interviews and narratives. The authoritative record of IEEE Spectrum's audio programming is the audio version.