Visualizing the medical data explosion
-
0:00 - 0:04I will start by posing a little bit of a challenge:
-
0:04 - 0:07the challenge of dealing with data,
-
0:07 - 0:09data that we have to deal with
-
0:09 - 0:11in medical situations.
-
0:11 - 0:13It's really a huge challenge for us.
-
0:13 - 0:15And this is our beast of burden --
-
0:15 - 0:17this is a Computer Tomography machine,
-
0:17 - 0:19a CT machine.
-
0:19 - 0:21It's a fantastic device.
-
0:21 - 0:23It uses X-rays, X-ray beams,
-
0:23 - 0:26that are rotating very fast around the human body.
-
0:26 - 0:28It takes about 30 seconds to go through the whole machine
-
0:28 - 0:30and is generating enormous amounts of information
-
0:30 - 0:32that comes out of the machine.
-
0:32 - 0:34So this is a fantastic machine
-
0:34 - 0:36that we can use
-
0:36 - 0:38for improving health care,
-
0:38 - 0:40but as I said, it's also a challenge for us.
-
0:40 - 0:43And the challenge is really found in this picture here.
-
0:43 - 0:45It's the medical data explosion
-
0:45 - 0:47that we're having right now.
-
0:47 - 0:49We're facing this problem.
-
0:49 - 0:51And let me step back in time.
-
0:51 - 0:54Let's go back a few years in time and see what happened back then.
-
0:54 - 0:56These machines that came out --
-
0:56 - 0:58they started coming in the 1970s --
-
0:58 - 1:00they would scan human bodies,
-
1:00 - 1:02and they would generate about 100 images
-
1:02 - 1:04of the human body.
-
1:04 - 1:06And I've taken the liberty, just for clarity,
-
1:06 - 1:09to translate that to data slices.
-
1:09 - 1:11That would correspond to about 50 megabytes of data,
-
1:11 - 1:13which is small
-
1:13 - 1:16when you think about the data we can handle today
-
1:16 - 1:18just on normal mobile devices.
-
1:18 - 1:20If you translate that to phone books,
-
1:20 - 1:23it's about one meter of phone books in the pile.
-
1:23 - 1:25Looking at what we're doing today
-
1:25 - 1:27with these machines that we have,
-
1:27 - 1:29we can, just in a few seconds,
-
1:29 - 1:31get 24,000 images out of a body,
-
1:31 - 1:34and that would correspond to about 20 gigabytes of data,
-
1:34 - 1:36or 800 phone books,
-
1:36 - 1:38and the pile would then be 200 meters of phone books.
-
1:38 - 1:40What's about to happen --
-
1:40 - 1:42and we're seeing this; it's beginning --
-
1:42 - 1:44a technology trend that's happening right now
-
1:44 - 1:47is that we're starting to look at time-resolved situations as well.
-
1:47 - 1:50So we're getting the dynamics out of the body as well.
-
1:50 - 1:52And just assume
-
1:52 - 1:55that we will be collecting data during five seconds,
-
1:55 - 1:57and that would correspond to one terabyte of data --
-
1:57 - 1:59that's 800,000 books
-
1:59 - 2:01and 16 kilometers of phone books.
-
2:01 - 2:03That's one patient, one data set.
-
2:03 - 2:05And this is what we have to deal with.
-
2:05 - 2:08So this is really the enormous challenge that we have.
-
2:08 - 2:11And already today -- this is 25,000 images.
-
2:11 - 2:13Imagine the days
-
2:13 - 2:15when we had radiologists doing this.
-
2:15 - 2:17They would put up 25,000 images,
-
2:17 - 2:20they would go like this, "25,0000, okay, okay.
-
2:20 - 2:22There is the problem."
-
2:22 - 2:24They can't do that anymore. That's impossible.
-
2:24 - 2:27So we have to do something that's a little bit more intelligent than doing this.
-
2:28 - 2:30So what we do is that we put all these slices together.
-
2:30 - 2:33Imagine that you slice your body in all these directions,
-
2:33 - 2:36and then you try to put the slices back together again
-
2:36 - 2:38into a pile of data, into a block of data.
-
2:38 - 2:40So this is really what we're doing.
-
2:40 - 2:43So this gigabyte or terabyte of data, we're putting it into this block.
-
2:43 - 2:45But of course, the block of data
-
2:45 - 2:47just contains the amount of X-ray
-
2:47 - 2:49that's been absorbed in each point in the human body.
-
2:49 - 2:51So what we need to do is to figure out a way
-
2:51 - 2:54of looking at the things we do want to look at
-
2:54 - 2:57and make things transparent that we don't want to look at.
-
2:57 - 2:59So transforming the data set
-
2:59 - 3:01into something that looks like this.
-
3:01 - 3:03And this is a challenge.
-
3:03 - 3:06This is a huge challenge for us to do that.
-
3:06 - 3:09Using computers, even though they're getting faster and better all the time,
-
3:09 - 3:11it's a challenge to deal with gigabytes of data,
-
3:11 - 3:13terabytes of data
-
3:13 - 3:15and extracting the relevant information.
-
3:15 - 3:17I want to look at the heart.
-
3:17 - 3:19I want to look at the blood vessels. I want to look at the liver.
-
3:19 - 3:21Maybe even find a tumor,
-
3:21 - 3:23in some cases.
-
3:24 - 3:26So this is where this little dear comes into play.
-
3:26 - 3:28This is my daughter.
-
3:28 - 3:30This is as of 9 a.m. this morning.
-
3:30 - 3:32She's playing a computer game.
-
3:32 - 3:34She's only two years old,
-
3:34 - 3:36and she's having a blast.
-
3:36 - 3:39So she's really the driving force
-
3:39 - 3:42behind the development of graphics-processing units.
-
3:43 - 3:45As long as kids are playing computer games,
-
3:45 - 3:47graphics is getting better and better and better.
-
3:47 - 3:49So please go back home, tell your kids to play more games,
-
3:49 - 3:51because that's what I need.
-
3:51 - 3:53So what's inside of this machine
-
3:53 - 3:55is what enables me to do the things that I'm doing
-
3:55 - 3:57with the medical data.
-
3:57 - 4:00So really what I'm doing is using these fantastic little devices.
-
4:00 - 4:02And you know, going back
-
4:02 - 4:04maybe 10 years in time
-
4:04 - 4:06when I got the funding
-
4:06 - 4:08to buy my first graphics computer --
-
4:08 - 4:10it was a huge machine.
-
4:10 - 4:13It was cabinets of processors and storage and everything.
-
4:13 - 4:16I paid about one million dollars for that machine.
-
4:17 - 4:20That machine is, today, about as fast as my iPhone.
-
4:22 - 4:24So every month there are new graphics cards coming out,
-
4:24 - 4:27and here is a few of the latest ones from the vendors --
-
4:27 - 4:30NVIDIA, ATI, Intel is out there as well.
-
4:30 - 4:32And you know, for a few hundred bucks
-
4:32 - 4:34you can get these things and put them into your computer,
-
4:34 - 4:37and you can do fantastic things with these graphics cards.
-
4:37 - 4:39So this is really what's enabling us
-
4:39 - 4:42to deal with the explosion of data in medicine,
-
4:42 - 4:44together with some really nifty work
-
4:44 - 4:46in terms of algorithms --
-
4:46 - 4:48compressing data,
-
4:48 - 4:51extracting the relevant information that people are doing research on.
-
4:51 - 4:54So I'm going to show you a few examples of what we can do.
-
4:54 - 4:57This is a data set that was captured using a CT scanner.
-
4:57 - 5:00You can see that this is a full data [set].
-
5:00 - 5:03It's a woman. You can see the hair.
-
5:03 - 5:06You can see the individual structures of the woman.
-
5:06 - 5:09You can see that there is [a] scattering of X-rays
-
5:09 - 5:11on the teeth, the metal in the teeth.
-
5:11 - 5:14That's where those artifacts are coming from.
-
5:14 - 5:16But fully interactively
-
5:16 - 5:19on standard graphics cards on a normal computer,
-
5:19 - 5:21I can just put in a clip plane.
-
5:21 - 5:23And of course all the data is inside,
-
5:23 - 5:26so I can start rotating, I can look at it from different angles,
-
5:26 - 5:29and I can see that this woman had a problem.
-
5:29 - 5:31She had a bleeding up in the brain,
-
5:31 - 5:33and that's been fixed with a little stent,
-
5:33 - 5:35a metal clamp that's tightening up the vessel.
-
5:35 - 5:37And just by changing the functions,
-
5:37 - 5:40then I can decide what's going to be transparent
-
5:40 - 5:42and what's going to be visible.
-
5:42 - 5:44I can look at the skull structure,
-
5:44 - 5:47and I can see that, okay, this is where they opened up the skull on this woman,
-
5:47 - 5:49and that's where they went in.
-
5:49 - 5:51So these are fantastic images.
-
5:51 - 5:53They're really high resolution,
-
5:53 - 5:55and they're really showing us what we can do
-
5:55 - 5:58with standard graphics cards today.
-
5:58 - 6:00Now we have really made use of this,
-
6:00 - 6:03and we have tried to squeeze a lot of data
-
6:03 - 6:05into the system.
-
6:05 - 6:07And one of the applications that we've been working on --
-
6:07 - 6:10and this has gotten a little bit of traction worldwide --
-
6:10 - 6:12is the application of virtual autopsies.
-
6:12 - 6:14So again, looking at very, very large data sets,
-
6:14 - 6:17and you saw those full-body scans that we can do.
-
6:17 - 6:20We're just pushing the body through the whole CT scanner,
-
6:20 - 6:23and just in a few seconds we can get a full-body data set.
-
6:23 - 6:25So this is from a virtual autopsy.
-
6:25 - 6:27And you can see how I'm gradually peeling off.
-
6:27 - 6:30First you saw the body bag that the body came in,
-
6:30 - 6:33then I'm peeling off the skin -- you can see the muscles --
-
6:33 - 6:36and eventually you can see the bone structure of this woman.
-
6:36 - 6:39Now at this point, I would also like to emphasize
-
6:39 - 6:41that, with the greatest respect
-
6:41 - 6:43for the people that I'm now going to show --
-
6:43 - 6:45I'm going to show you a few cases of virtual autopsies --
-
6:45 - 6:47so it's with great respect for the people
-
6:47 - 6:49that have died under violent circumstances
-
6:49 - 6:52that I'm showing these pictures to you.
-
6:53 - 6:55In the forensic case --
-
6:55 - 6:57and this is something
-
6:57 - 6:59that ... there's been approximately 400 cases so far
-
6:59 - 7:01just in the part of Sweden that I come from
-
7:01 - 7:03that has been undergoing virtual autopsies
-
7:03 - 7:05in the past four years.
-
7:05 - 7:08So this will be the typical workflow situation.
-
7:08 - 7:10The police will decide --
-
7:10 - 7:12in the evening, when there's a case coming in --
-
7:12 - 7:15they will decide, okay, is this a case where we need to do an autopsy?
-
7:15 - 7:18So in the morning, in between six and seven in the morning,
-
7:18 - 7:20the body is then transported inside of the body bag
-
7:20 - 7:22to our center
-
7:22 - 7:24and is being scanned through one of the CT scanners.
-
7:24 - 7:26And then the radiologist, together with the pathologist
-
7:26 - 7:28and sometimes the forensic scientist,
-
7:28 - 7:30looks at the data that's coming out,
-
7:30 - 7:32and they have a joint session.
-
7:32 - 7:35And then they decide what to do in the real physical autopsy after that.
-
7:37 - 7:39Now looking at a few cases,
-
7:39 - 7:41here's one of the first cases that we had.
-
7:41 - 7:44You can really see the details of the data set.
-
7:44 - 7:46It's very high-resolution,
-
7:46 - 7:48and it's our algorithms that allow us
-
7:48 - 7:50to zoom in on all the details.
-
7:50 - 7:52And again, it's fully interactive,
-
7:52 - 7:54so you can rotate and you can look at things in real time
-
7:54 - 7:56on these systems here.
-
7:56 - 7:58Without saying too much about this case,
-
7:58 - 8:00this is a traffic accident,
-
8:00 - 8:02a drunk driver hit a woman.
-
8:02 - 8:05And it's very, very easy to see the damages on the bone structure.
-
8:05 - 8:08And the cause of death is the broken neck.
-
8:08 - 8:10And this women also ended up under the car,
-
8:10 - 8:12so she's quite badly beaten up
-
8:12 - 8:14by this injury.
-
8:14 - 8:17Here's another case, a knifing.
-
8:17 - 8:19And this is also again showing us what we can do.
-
8:19 - 8:21It's very easy to look at metal artifacts
-
8:21 - 8:24that we can show inside of the body.
-
8:24 - 8:27You can also see some of the artifacts from the teeth --
-
8:27 - 8:29that's actually the filling of the teeth --
-
8:29 - 8:32but because I've set the functions to show me metal
-
8:32 - 8:34and make everything else transparent.
-
8:34 - 8:37Here's another violent case. This really didn't kill the person.
-
8:37 - 8:39The person was killed by stabs in the heart,
-
8:39 - 8:41but they just deposited the knife
-
8:41 - 8:43by putting it through one of the eyeballs.
-
8:43 - 8:45Here's another case.
-
8:45 - 8:47It's very interesting for us
-
8:47 - 8:49to be able to look at things like knife stabbings.
-
8:49 - 8:52Here you can see that knife went through the heart.
-
8:52 - 8:54It's very easy to see how air has been leaking
-
8:54 - 8:56from one part to another part,
-
8:56 - 8:59which is difficult to do in a normal, standard, physical autopsy.
-
8:59 - 9:01So it really, really helps
-
9:01 - 9:03the criminal investigation
-
9:03 - 9:05to establish the cause of death,
-
9:05 - 9:08and in some cases also directing the investigation in the right direction
-
9:08 - 9:10to find out who the killer really was.
-
9:10 - 9:12Here's another case that I think is interesting.
-
9:12 - 9:14Here you can see a bullet
-
9:14 - 9:17that has lodged just next to the spine on this person.
-
9:17 - 9:20And what we've done is that we've turned the bullet into a light source,
-
9:20 - 9:22so that bullet is actually shining,
-
9:22 - 9:25and it makes it really easy to find these fragments.
-
9:25 - 9:27During a physical autopsy,
-
9:27 - 9:29if you actually have to dig through the body to find these fragments,
-
9:29 - 9:31that's actually quite hard to do.
-
9:33 - 9:35One of the things that I'm really, really happy
-
9:35 - 9:38to be able to show you here today
-
9:38 - 9:40is our virtual autopsy table.
-
9:40 - 9:42It's a touch device that we have developed
-
9:42 - 9:45based on these algorithms, using standard graphics GPUs.
-
9:45 - 9:47It actually looks like this,
-
9:47 - 9:50just to give you a feeling for what it looks like.
-
9:50 - 9:53It really just works like a huge iPhone.
-
9:53 - 9:55So we've implemented
-
9:55 - 9:58all the gestures you can do on the table,
-
9:58 - 10:02and you can think of it as an enormous touch interface.
-
10:02 - 10:04So if you were thinking of buying an iPad,
-
10:04 - 10:07forget about it. This is what you want instead.
-
10:07 - 10:10Steve, I hope you're listening to this, all right.
-
10:11 - 10:13So it's a very nice little device.
-
10:13 - 10:15So if you have the opportunity, please try it out.
-
10:15 - 10:18It's really a hands-on experience.
-
10:18 - 10:21So it gained some traction, and we're trying to roll this out
-
10:21 - 10:23and trying to use it for educational purposes,
-
10:23 - 10:25but also, perhaps in the future,
-
10:25 - 10:28in a more clinical situation.
-
10:28 - 10:30There's a YouTube video that you can download and look at this,
-
10:30 - 10:32if you want to convey the information to other people
-
10:32 - 10:35about virtual autopsies.
-
10:35 - 10:37Okay, now that we're talking about touch,
-
10:37 - 10:39let me move on to really "touching" data.
-
10:39 - 10:41And this is a bit of science fiction now,
-
10:41 - 10:44so we're moving into really the future.
-
10:44 - 10:47This is not really what the medical doctors are using right now,
-
10:47 - 10:49but I hope they will in the future.
-
10:49 - 10:52So what you're seeing on the left is a touch device.
-
10:52 - 10:54It's a little mechanical pen
-
10:54 - 10:57that has very, very fast step motors inside of the pen.
-
10:57 - 10:59And so I can generate a force feedback.
-
10:59 - 11:01So when I virtually touch data,
-
11:01 - 11:04it will generate forces in the pen, so I get a feedback.
-
11:04 - 11:06So in this particular situation,
-
11:06 - 11:08it's a scan of a living person.
-
11:08 - 11:11I have this pen, and I look at the data,
-
11:11 - 11:13and I move the pen towards the head,
-
11:13 - 11:15and all of a sudden I feel resistance.
-
11:15 - 11:17So I can feel the skin.
-
11:17 - 11:19If I push a little bit harder, I'll go through the skin,
-
11:19 - 11:22and I can feel the bone structure inside.
-
11:22 - 11:24If I push even harder, I'll go through the bone structure,
-
11:24 - 11:27especially close to the ear where the bone is very soft.
-
11:27 - 11:30And then I can feel the brain inside, and this will be the slushy like this.
-
11:30 - 11:32So this is really nice.
-
11:32 - 11:35And to take that even further, this is a heart.
-
11:35 - 11:38And this is also due to these fantastic new scanners,
-
11:38 - 11:40that just in 0.3 seconds,
-
11:40 - 11:42I can scan the whole heart,
-
11:42 - 11:44and I can do that with time resolution.
-
11:44 - 11:46So just looking at this heart,
-
11:46 - 11:48I can play back a video here.
-
11:48 - 11:50And this is Karljohan, one of my graduate students
-
11:50 - 11:52who's been working on this project.
-
11:52 - 11:55And he's sitting there in front of the Haptic device, the force feedback system,
-
11:55 - 11:58and he's moving his pen towards the heart,
-
11:58 - 12:00and the heart is now beating in front of him,
-
12:00 - 12:02so he can see how the heart is beating.
-
12:02 - 12:04He's taken the pen, and he's moving it towards the heart,
-
12:04 - 12:06and he's putting it on the heart,
-
12:06 - 12:09and then he feels the heartbeats from the real living patient.
-
12:09 - 12:11Then he can examine how the heart is moving.
-
12:11 - 12:13He can go inside, push inside of the heart,
-
12:13 - 12:16and really feel how the valves are moving.
-
12:16 - 12:19And this, I think, is really the future for heart surgeons.
-
12:19 - 12:22I mean it's probably the wet dream for a heart surgeon
-
12:22 - 12:25to be able to go inside of the patient's heart
-
12:25 - 12:27before you actually do surgery,
-
12:27 - 12:29and do that with high-quality resolution data.
-
12:29 - 12:31So this is really neat.
-
12:32 - 12:35Now we're going even further into science fiction.
-
12:35 - 12:38And we heard a little bit about functional MRI.
-
12:38 - 12:41Now this is really an interesting project.
-
12:41 - 12:43MRI is using magnetic fields
-
12:43 - 12:45and radio frequencies
-
12:45 - 12:48to scan the brain, or any part of the body.
-
12:48 - 12:50So what we're really getting out of this
-
12:50 - 12:52is information of the structure of the brain,
-
12:52 - 12:54but we can also measure the difference
-
12:54 - 12:57in magnetic properties of blood that's oxygenated
-
12:57 - 13:00and blood that's depleted of oxygen.
-
13:00 - 13:02That means that it's possible
-
13:02 - 13:04to map out the activity of the brain.
-
13:04 - 13:06So this is something that we've been working on.
-
13:06 - 13:09And you just saw Motts the research engineer, there,
-
13:09 - 13:11going into the MRI system,
-
13:11 - 13:13and he was wearing goggles.
-
13:13 - 13:15So he could actually see things in the goggles.
-
13:15 - 13:18So I could present things to him while he's in the scanner.
-
13:18 - 13:20And this is a little bit freaky,
-
13:20 - 13:22because what Motts is seeing is actually this.
-
13:22 - 13:25He's seeing his own brain.
-
13:25 - 13:27So Motts is doing something here,
-
13:27 - 13:29and probably he is going like this with his right hand,
-
13:29 - 13:31because the left side is activated
-
13:31 - 13:33on the motor cortex.
-
13:33 - 13:35And then he can see that at the same time.
-
13:35 - 13:37These visualizations are brand new.
-
13:37 - 13:40And this is something that we've been researching for a little while.
-
13:40 - 13:43This is another sequence of Motts' brain.
-
13:43 - 13:46And here we asked Motts to calculate backwards from 100.
-
13:46 - 13:48So he's going "100, 97, 94."
-
13:48 - 13:50And then he's going backwards.
-
13:50 - 13:53And you can see how the little math processor is working up here in his brain
-
13:53 - 13:55and is lighting up the whole brain.
-
13:55 - 13:57Well this is fantastic. We can do this in real time.
-
13:57 - 13:59We can investigate things. We can tell him to do things.
-
13:59 - 14:01You can also see that his visual cortex
-
14:01 - 14:03is activated in the back of the head,
-
14:03 - 14:05because that's where he's seeing, he's seeing his own brain.
-
14:05 - 14:07And he's also hearing our instructions
-
14:07 - 14:09when we tell him to do things.
-
14:09 - 14:11The signal is really deep inside of the brain as well,
-
14:11 - 14:13and it's shining through,
-
14:13 - 14:15because all of the data is inside this volume.
-
14:15 - 14:17And in just a second here you will see --
-
14:17 - 14:19okay, here. Motts, now move your left foot.
-
14:19 - 14:21So he's going like this.
-
14:21 - 14:23For 20 seconds he's going like that,
-
14:23 - 14:25and all of a sudden it lights up up here.
-
14:25 - 14:27So we've got motor cortex activation up there.
-
14:27 - 14:29So this is really, really nice,
-
14:29 - 14:31and I think this is a great tool.
-
14:31 - 14:33And connecting also with the previous talk here,
-
14:33 - 14:35this is something that we could use as a tool
-
14:35 - 14:37to really understand
-
14:37 - 14:39how the neurons are working, how the brain is working,
-
14:39 - 14:42and we can do this with very, very high visual quality
-
14:42 - 14:45and very fast resolution.
-
14:45 - 14:47Now we're also having a bit of fun at the center.
-
14:47 - 14:50So this is a CAT scan -- Computer Aided Tomography.
-
14:51 - 14:53So this is a lion from the local zoo
-
14:53 - 14:56outside of Norrkoping in Kolmarden, Elsa.
-
14:56 - 14:58So she came to the center,
-
14:58 - 15:00and they sedated her
-
15:00 - 15:02and then put her straight into the scanner.
-
15:02 - 15:05And then, of course, I get the whole data set from the lion.
-
15:05 - 15:07And I can do very nice images like this.
-
15:07 - 15:09I can peel off the layer of the lion.
-
15:09 - 15:11I can look inside of it.
-
15:11 - 15:13And we've been experimenting with this.
-
15:13 - 15:15And I think this is a great application
-
15:15 - 15:17for the future of this technology,
-
15:17 - 15:20because there's very little known about the animal anatomy.
-
15:20 - 15:23What's known out there for veterinarians is kind of basic information.
-
15:23 - 15:25We can scan all sorts of things,
-
15:25 - 15:27all sorts of animals.
-
15:27 - 15:30The only problem is to fit it into the machine.
-
15:30 - 15:32So here's a bear.
-
15:32 - 15:34It was kind of hard to get it in.
-
15:34 - 15:37And the bear is a cuddly, friendly animal.
-
15:37 - 15:40And here it is. Here is the nose of the bear.
-
15:40 - 15:43And you might want to cuddle this one,
-
15:43 - 15:46until you change the functions and look at this.
-
15:46 - 15:48So be aware of the bear.
-
15:48 - 15:50So with that,
-
15:50 - 15:52I'd like to thank all the people
-
15:52 - 15:54who have helped me to generate these images.
-
15:54 - 15:56It's a huge effort that goes into doing this,
-
15:56 - 15:59gathering the data and developing the algorithms,
-
15:59 - 16:01writing all the software.
-
16:01 - 16:04So, some very talented people.
-
16:04 - 16:07My motto is always, I only hire people that are smarter than I am
-
16:07 - 16:09and most of these are smarter than I am.
-
16:09 - 16:11So thank you very much.
-
16:11 - 16:15(Applause)
- Title:
- Visualizing the medical data explosion
- Speaker:
- Anders Ynnerman
- Description:
-
Today medical scans produce thousands of images and terabytes of data for a single patient in mere seconds, but how do doctors parse this information and determine what's useful? At TEDxGöteborg, scientific visualization expert Anders Ynnerman shows us sophisticated new tools -- like virtual autopsies -- for analyzing this myriad data, and a glimpse at some sci-fi-sounding medical technologies in development. This talk contains some graphic medical imagery.
- Video Language:
- English
- Team:
closed TED
- Project:
- TEDTalks
- Duration:
- 16:16
![]() |
TED edited English subtitles for Visualizing the medical data explosion | |
![]() |
TED added a translation |