Daniel Keefe For Keefe, designing new ways to visualize data is almost like making art.
4:03 PM me: hello Daniel: hi me: ready for our chat?
Daniel: yes
4:05 PM me: great. so first, where are you chatting from?
Daniel: I'm in Minneapolis, in my office at the U. of Minnesota. me: What's the view from your window? Daniel:
Sunny day, a tree blowin' in the wind.. I'm looking out over a grassy
area in front of one of the engineering buildings, and lots of bikers.
This is a great city for biking.
4:08 PM me: do you bike to the university?
Daniel: I do often
me: do you think about your research while riding to work? Daniel:
I prefer to imagine I'm riding on Le Tour. There's a great path here that
goes right along the Mississippi River, I usually take that in to work.
4:12 PM me: Sounds really lovely.
So let's talk about your research now. I know that you're working on 3D visualizations of animal movement. Can you first talk about why you're doing this?
4:15 PM Daniel:
I'm working with an interdisciplinary group of researchers
(evolutionary biologists, orthopedists, computer scientists). The
biologists are the driving force behind the animal movement studies.
They are interested in studying interesting movements in animals today
and from that understanding more about these animals and potentially
understanding something about how certain motions and anatomical
features evolved over time. me: Ok. and your role in this?
4:17 PM Daniel:
I'm interested in these projects because the data involved is quite
complicated. My research centers around learning how we can better use
computer technology to visualize and explore complicated scientific
data. me: Why are you interested in using your computer skills to this end?
4:24 PM Daniel:
I enjoy the process. There are many opportunities to be creative..
designing new ways to visualize data, it's almost like making art. It
requires a real visual investigation. I also believe it's important
work. If we can better understand how to use computer tools to
visualize complex motions in animals, then that technology can help us
to study disease and treatments in humans in a number of contexts.
4:26 PM me: do you any background or experience as an artist? Daniel: Yes, I've studied art, and I do quite a bit of painting.
4:29 PM me: Interesting so how does your art training come into play? What did you mean by a visual investigation?
4:36 PM Daniel:
Well, the goal in data visualization is to develop a visual
representation for the data that people can understand. In the simplest
case, this might be a bar chart, most people are familiar with that.
But, what happens if your data are much more complicated, for example,
let's say you're a doctor working with a 3D CT scan. How do we make
that data visual?
That turns out to not be too difficult using computers, but let's up the complexity even more. Now let's say you want to see how the bones that show up in a CT scan move while your patient is performing some action. Now, we need a 3D animated picture, so it's getting more complicated. We can keep going, we may need to know some information about the forces exerted on the bones, or the minimum distance of one bone to another. So, the data get more and more complex this way, and as it gets more complex, coming up with a visual to describe the situation is more and more difficult. The process we go through to develop those visuals, is quite similar to an artist's process. There is a lot of design and critique and refinement that goes into it.. a lot of thought about color and form and narrative and metaphor.. these are all things that artists think about quite often. 4:40 PM me:
What are some of the data points that the team is collecting that you
need to convert into an animation? How is it different from taking a
movie of an animal and looking at that?
4:44 PM Daniel:
We do use movies, but they are special.. right now, we're collecting
two high-speed X-ray movies of the animal, each from a different angle.
We also take a CT scan of the animal. From the CT scan, we learn the
shape of the bones of the animal, and from the X-ray movies, we learn
how the bones move. Then, we use computers to make an interactive 3D
movie of the bones moving that you can view in virtual reality.In
this virtual reality environment, you can look at the skeleton in
motion from any angle and control the playback, pausing, rewinding,
etc. You can also calculate axes of rotation for the bones and plot
data about the movement of the bones. So, it's much more informative
than a regular movie.
me: So what does the 3D movie reveal?
4:51 PM Daniel:
One of the specific examples we've been looking at is a pig chewing. It
turns out to be a very interesting motion to study. You've heard about
people who have had problems with their TMJ -- it's the joint where
your skull and jaw attach. The mechanics of this joint are very
interesting, especially in pigs. Through these visualizations we've
been learning more about how this joint works in pigs.
We're trying to determine now how the motion of this joint is different in pigs than it is in humans --- it looks to be quite different! Another thing we're interested in is how the motion in the joint changes depending upon the amount of food in the mouth. The 3D visualizations are constructed to describe this motion as clearly as possible, so the scientists can interpret it. 4:53 PM me: Do you do some of the same visualizations on people? To look at how they chew? Daniel:
We haven't yet, but in developing ideas for the visualizations, we have
built upon research that has been published in human dentistry. I'm
sure our techniques would be quite helpful in looking at human chewing
patterns. We have some projects that are just starting now to look at
other human motions, for example motion of the spine and the wrist.
4:57 PM me: Okay, but what I don't understand is how does looking at pig chewing tell you something about how humans chew?
5:01 PM Daniel:
It doesn't necessarily. The question I'm trying to tackle is, for
humans, pigs, or anything else.. how do we visualize motions
(biomechanics) using computers so that we can understand them. Doctors
and scientists will have to figure out for themselves what conclusions
to draw from the data, but right now, they can't even see their data.
I'm trying to develop new ways for them to see their data.
5:03 PM me:
I see. So can you give me an example of a visualization challenge that
you've been able to overcome? Or maybe one that you are still have
problems with?
5:08 PM Daniel:
Sure, one we've been focusing on recently is describing how the axis of
rotation for the pig's jaw changes over time. So, if this jaw were a
simple hinge like a door (which it's not) then the axis of rotation
would go right through the joint where the jaw and the skull attach.
The interesting thing is that the joint is not a simple hinge, there's sliding and rotation that happens at different points in the chewing cycle as the jaw opens and the food is ground up. We've created visualizations that calculate the correct axis of rotation for each instant in time and then display this information so the scientists can see how the location of the axis changes depending upon where the pig is in the chewing cycle. Based upon this, it's very clear that the jaw is not acting as a simple hinge. 5:10 PM me: When you see the data transformed into the 3D viz, how does that make you feel?
5:12 PM Daniel:
It looks pretty funky! This pig skull could easily pass for dinosaur
bones... I mean it just looks exciting. Then to see it move and chew is
fascinating. You know, I usually look at it in slow motion because it's
so complex, it's easier to understand that way. But, when you view it
at normal speed, it's amazing how fast these guys are chewing. I mean
they just tear through the food. So, there really is some basis behind
the saying, "You're eating like a pig!"
5:15 PM me: ha!
nice
Would you like to analyze other animals like this?
5:20 PM Daniel:
Yes, we can learn a lot from studying the motions of animals.
Sometimes, it provides clues about evolution. We've also worked quite a
bit with bats. The way that bats fly is completely different from our
airplanes. We can learn a lot about flight from bats. So, there's
plenty to learn here from a number of animals, and I think these
visualization techniques can be helpful in many of these cases.
5:21 PM me:
I know our time is running short and I don't want to keep you too much
longer, but can you give an example of how the motions may provide
clues about evolution?
5:28 PM Daniel:
One of my favorite examples has to do with studying pigs walking. As I
understand, pigs have a particular tendon in their leg now, and if you
trace back through the hypothesized evolutionary history, you'll find
examples of animals that are ancestors of pigs which have a muscle in
the place of this tendon. In other words, at some point, it seems that
this tendon stopped being muscle and started being tendon.
So, one question is, why did that happen? If we can understand the exact role of the tendon in the movement of the pig's leg today, then we should get some insight into understanding why the tendon feature may have evolved in the first place and through that we can learn more about animals that are now extinct. 5:30 PM me: Very interesting. Well,
I suppose I should wrap this up. I'll be looking forward to the images
you send.
Daniel: Sure, I'll send that along in email
5:32 PM me: Great. So what are you going to do with the rest of your afternoon?
5:33 PM Daniel: I'm going to get back to reading some papers here.. all about processing motion data of this sort.
me: Ok, then have a nice day.
5:34 PM Daniel: Thanks, you too. |
advertisement
What's On Now
|
