UCLA Develops World's Fastest Camera To Hunt Down Cancer In Real Time

MrSeb writes "Engineers at UCLA, led by Bahram Jalali and Dino Di Carlo, have developed a camera that can take 36.7 million frames per second, with a shutter speed of 27 picoseconds. By far the fastest and most sensitive camera in the world — it is some 100 times faster than existing optical microscopes, and it has a false-positive rate of just one in a million — it is hoped, among other applications, that the device will massively improve our ability to diagnose early-stage and pre-metastatic cancer. This camera can photograph single cells as they flow through a microfluidic system at four meters per second (9 mph — about 100,000 particles per second), with comparable image quality to a still CCD camera (with a max shooting speed of around 60 fps). Existing optical microscopes use CMOS sensors, but they're not fast enough to image more than 1,000 particles per second. With training, the brains of the operation — an FPGA image processor — can automatically analyze 100,000 particles per second and detect rare particles (such as cancer cells) 75% of the time."

Soon:

[Anne_Nonymous]

LAPD uses fastest camera to record traffic violations before they occur.

Oh just wait

[kiriath]

Wait until Mythbusters has their hands on one of THESE. An entire show comprised of one bullet hitting a target? Much Win.

Re:Oh just wait

[durrr]

At 24fps playback, light itself would move only 192 meters per second. A pistol bullet would move at less than 1 mm a second resulting in a pretty much still-footage look.

Blinking the lights of a football stadium would look nice though.

Re:

[loustic]

Can't wait to see the pr0n industry taking advantage of this new piece of technology. Available in 4k, 36.7Mfps and 3D, on demand !

Re:

[PIBM]

Sadly, this detector only gather 36.7 mPixels/second since it has a single photodetector. In comparison, at the UK olympics, videos will be filmed and displayed in 7680 x 4320 at 120 fps, a total of 3981.3 mPixels/second, or 100 times more than their 100 times faster camera!

Re:

[K. S. Kyosuke]

Can't wait to see the pr0n industry taking advantage of this new piece of technology. Available in 4k, 36.7Mfps and 3D, on demand !

Well, if you get off on bacterial conjugation porn, this would be a nice piece of hardware to film it, but the last time I checked this particular market was pretty small.

Re:

[DragonTHC]

that's nothing, wait until porn directors gets one of these!

Re:

[Rogerborg]

Still not a high enough frame rate to track Kari Byron's polychromatic hair in real time though.

Re:

[Anonymous Coward]

The more likely scenario is the camera will be taken to the bomb range, packed with black powder and blown up.

Re:

[Greyfox]

Hell with that! If they get one of these, I want to see them detonate a nuke!

Re:

[kermidge]

This is not the link I was looking for, which showed detonation _inside_ the device casing, but it's a start:

http://www.waynesthisandthat.com/abombs.html [waynesthisandthat.com] while

https://en.wikipedia.org/wiki/High-speed_photography [wikipedia.org] gives some good background reading, references, and links.

What the folks at UCLA have done seems pretty neat for getting real-world results (but maybe I'm easily amused.)

100 times faster than existing optical microscopes

[vlm]

100 times faster than existing optical microscopes

What does that mean before it went thru the journalist filter?

I haven't done film photography in twenty years, but I still thought of "fast lenses" like those weirdo F/0.3 "lenses" made outta mirrors and pretty much anything lower than F/1. You know, the lenses that cost about as much as a cheap new car. Yeah, a 35 picosecond shutter speed means you need a pretty fast lens to feed enough photons thru it, but what does that factor of 100 actually mean, like F/0.001 lens exists or something?

Re:

[Infiniti2000]

Lens speed has nothing to do with this. This is not a normal camera. They fire laser pulses through lenses with a static aperture and the reflection from these laser pulses is what is processed to generate the image.

Re:

[mcgrew]

I haven't done film photography in twenty years

I can tell. The lens and F stop have nothing to do with speed, the shutter does. Of course with fast shutter speed you need fast film (is there still any around?) and a larger aperture (F-stop).

The lens has nothing to do with a camera's speed.

Re:

[Tool Man]

I can tell. The lens and F stop have nothing to do with speed, the shutter does. Of course with fast shutter speed you need fast film (is there still any around?) and a larger aperture (F-stop).

The lens has nothing to do with a camera's speed.

The lens has everything to do with capturing enough light to make a useful image within that short shutter time. I keep my f1.4 prime 50mm on my Pentax, because it does such a nice job without a flash much of the time.

Re:100 times faster than existing optical microsco

[vlm]

The lens and F stop have nothing to do with speed, the shutter does.

Nah there are fast lenses and slow lenses. At least that's what they called them in the 90s. A fast lens can run a faster shutter speed usually because its bigger but also "insert optical magic here".

Just like you can have two telescopes with identical focus and identical field of view but the one with twice the lens surface area has a shutter speed twice as fast, because it shovels in twice as many photons per second.

Think how easy it is to make a zoom lens for sunlit outdoor work, but how hard it is to pull off a long distance candle-lit shot... you can do it with a fast lens that basically resembles a large aperture telescope, or theres some heavy optical weirdness with trading off the depth of field for light strength or something.

I know just enough optics to be really dangerous. Mostly because I know microwave RF optics better than I know visual optics.

Here I found the lens speed article on wikipedia:

http://en.wikipedia.org/wiki/Lens_speed [wikipedia.org]

Re:

[awrowe]

I think you will find a lens which has a low F ratio is known as a "fast" lens.

Looking at a telescope for example, my 10inch Newtonian Reflector has a focal length of 1200mm and an aperture of 254mm, which ends up giving it an f ratio of 4.7, which is considered reasonably "fast" in optical terms. By comparison, a long tubed refractor may have an f ratio of 15, which would make it "slow". The "fast" or "slow" is referring to the amount of time it takes to get x amount of photons to the eyeball/imaging devic

Re:100 times faster than existing optical microsco

[tlhIngan]

I can tell. The lens and F stop have nothing to do with speed, the shutter does. Of course with fast shutter speed you need fast film (is there still any around?) and a larger aperture (F-stop).

The lens has nothing to do with a camera's speed.

And you're speaking nonsense.

First, there are three variables at play when taking a photo. The shutter speed, the aperture size (usually expressed as a fraction of focal length, f), and the "film" speed (sensitivity).

A properly exposed photo (usually desired in most cases excepting artistic composition) is where sufficient photons hit the light-sensitive medium to generate a usable image. With 3 variables to play with, there is a whole range of settings that will generate an image.

Hence the concepts of stuff like "stops" - where you can halve the shutter speed (1/30th from 1/60th, say), but also close down the aperture (by half) and still end up with a good exposure. Or decrease the media's sensitivity.

What you use depends on the situation and artistic effects you want to put on it - larger apertures reduce depth of field, faster shutters "freeze" fast motion, more sensitive recording media generally are noisier/grainier, etc.

A "fast" lens is called that because it has a very wide aperture that can let in lots of light, meaning you can get very nicely exposed photos at very high shutter speeds, thus capturing faster motion. A "slow" lens means the aperture is smaller and thus requires a longer exposure time (slower shutter speed).

Remember, it's all about getting the proper exposure, and playing with the three variables will generate many configurations of apeture/shutter/sensitivity that will work. The one you use depends on the situation.

Re:

[mug funky]

wtf you talking about? you ever met a photographer?

when a photographer colloquially refers to a "fast" lens, they are saying the lens lets through comparatively more light than an "average lens". this enables them to shoot with a higher shutter speed, hence the lens enables a faster shot for the same film rating.

i suppose they could call them "light" and "dark" lenses and it'd make more sense from an optical point of view, but in practical terms when shooting in low light, you want fast film, fast lens an

Re:

[sFurbo]

As sibling post said, it is not about the lens, or even the shutter, as the article says. Think of of it as a very fast flash in a dark room. I have no idea what they are talking about with the optical microscope, I don't see how they should have an inherent speed.

Re:100 times faster than existing optical microsco

[durrr]

Don't think of lenses at all, this is a specialist setup made to picture cells and is more something along the line of the bastard child of a flow cytometer and a picosecond laser system than anything resembling classical photography equipment.

Re:

[vlm]

Its a LIDAR ! More or less?

Re:

[dmearns]

Comparing the camera shutter speed to an optical microscope makes about as much sense as saying the camera "has a false-positive rate of just one in a million". Someone forgot to read the article before posting, again.

Re:

[pmathew]

It may be referring to the optical time stretching analog to digital converter(ADC) which i guess is used in the system (http://en.wikipedia.org/wiki/Time_stretch_analog-to-digital_converter) . All those optical stuff shown in the video is for that purpose is my guess . Since shutter action is like a sampling event we need to convert these sampled data into digital at extreme data rate 100s of GSPS which is vey challenging and is done by TSADC using time stretching principle and then using several interleav

Re:

[chilenexus]

Likewise, how can a camera have a 1 in 1,000,000 false positive rate? That makes as much sense as "how many miles per gallon does your phone get?" This is probably the author including the processing software that analyzes the pictures taken by the camera?

storage headache

[airuck]

As anyone supporting nextgen sequencing will know, researchers will claim that they simply must have the raw data. My dissuading answer: OK, then THIS is how much it is going to cost you. Payment due in full before data acquisition.

Still not fast enough

[andyring]

Even at 36.7 million FPS, it's still too slow to freeze-frame a Chuck Norris roundhouse kick...

Unfair attributions

[Anonymous Coward]

UCLA's original press release can be read at http://newsroom.ucla.edu/portal/ucla/world-s-fastest-camera-used-to-235979.aspx [ucla.edu], which also indicates that the research team was led by Bahram Jalali and Dino Di Carlo. However, the team consisted of Keisuke Goda, Ali Ayazi, Daniel R. Gossett, Jagannath Sadasivam, Cejo K. Lonappan, Elodie Sollier, Ali M. Fard, Soojung Claire Hur, Jost Adam, Coleman Murray, Chao Wang, Nora Brackbill, Dino Di Carlo, and Bahram Jalali, who contributed the following to the work: K.G

So many potential uses beyond medicine

[jenningsthecat]

It's awesome what this can do for early cancer detection and the like, but just think of all the things beyond healthcare that it can be used for.

I could wee this combined with nanobots to do on-the-fly water purification, or even a kind of fractional distillation. Not to mention the possibilities for industrial process control, food, pharmaceuticals, and on and on.

Here it comes

[Sparticus789]

Create a video game in which the user flies through the body with a micro-fighter purging the body of cancer cells. There micro-fighters are real fighters that are inside someone's body, and this camera provides the real-time images. Then whoever wins the game gets a scholarship to become a doctor, healing casualties of an intergalactic war. The game winner also gets a robot to take their place on Earth

AND

Uwe Boll turns this into a terrible movie.

Re:

[heathen_01]

Have you been reading enders game again?

False positive what?

[wonkey_monkey]

and it has a false-positive rate of just one in a million

Having skimmed the article - naughty me - it seems like it's the already existing FPGA that has a false-positive rate (for detecting cancer, which hadn't been mentioned in the text up to this point) of 1/1000000, and they were just waiting for a fast enough camera to come along.

I cannot wait...

[wbr1]

...till someobe starts filmong porn with it. It will be as good as IMAX porn where one razor bump on a labia is bigger than your face.

Failure rate * fps

[PPalmgren]

Umm, I'm a bit confused. It takes almost 40 million frames per second and has a one in a million error rate? Is this a good thing or a bad thing being spun as good? Anyone care to elaborate?

Re:

[Sardak]

I read it as one in a million cells are falsely identified. So, every 10 seconds or so (at 100,000/sec) it would incorrectly classify one. Sounds like a bad thing being spun as good.

Re:

[canajin56]

The expected number of false positives per is 10-20 per mL of blood, while the expected number of true positives depends on concentrations of the cancerous cells. So basically you put a 10 mL sample in, and 15 minutes later you have 100+ pictures. If the cells are not present, all 100-200ish will be white blood cells, but if it cells are present, half of the pictures will be of the target cell, and that's if the cells are occurring about one out of every million. At any rate, a human expert is needed to

wrong calculation

[mtmmtm]

Sorry. You have done a wrong calculation it should be 27 nanoseconds.

Dialysis-like treatment?

[Arpie]

Not medically savvy here, but could this lead to some sort of Dialysis-like process to remove cancer cells from the bloodstream? I guess it would help with at least some types of cancer?

1 Trillion FPS

[White Flame]

MIT hit a lot faster than a measly 36.7 million fps.

http://web.media.mit.edu/~raskar/trillionfps/ [mit.edu]

Motion blur

[jovius]

It's impossible to have sharp images of the universe, because in the end the camera itself is not precisely defined. It's interesting that technology can never be at par with the local physical conditions - we will always be a tiny fraction of a fraction behind. There can't exist an infinite frame rate device. In other words - if one shoots a video there always is change between the frames.

What about the Femto Photography project at MIT?

[fddr]

This does not sound like the fastest camera in the world to me. I believe the Femto Photography project within the Camera Culture group at the MIT Media Lab (http://web.media.mit.edu/~raskar/trillionfps/) deals with frame rates that are approximately 5 orders of magnitude greater than reported in this article.