T. E. R. Phillips

Transcription

I believe that we can both unravel the mysteries of the universe, and save human lives at the same time through interdisciplinary research. And I'm gonna share with you today just one story, my story, that has crossed these paths. And we start the in supernova remnant Casiopea A, it's one of the youngest ones in our galaxy, about 330 years old. An astronomy colleague approached me one day. And she had over 8 years of magnificent data, just trying to understand the 3D structure of this nebula, the supernova remnant, but she had no way to look at it. So I looked at the data with her and said, "I think I can help you." And although -- and this is all real data you're seeing on the screen above me -- this is the Hollywood rendering version, but the rough draft I made with her, looks something more like this. And she was able to make novel discoveries about how supernovas explode, and how shells explode within it, using a piece of software developed at Brigham and Women's Hospital, here in Boston, called "3D Slicer", and it was originally developed for looking at patients' brain scans during surgical planning and doing 3D renderings of anatomy. Who knew our solution was lurking just across the river. Now, people don't believe me when I tell them that astronomy and medical imaging, these two seemingly different fields, are really similar. So we're gonna play a little game I like to call "Which is which." I play this with new doctors and astronomers I work with. I'm gonna show you two images on the screen. One of them is biomedical and one of them is astronomical and you have to pick them correctly in your head. So, here is the first set. And again, one of these is a biomedical and one is astronomical. I'll give you a second to make your little vote mentally -- So it turns out the one on the left is some of the raw data of the supernova remnant we were just looking at, and on the right we have an angiogram of a patient's heart and coronary arteries. Okay, we're gonna try another one. Now, this one is much closer to my daily bread and butter. Now, tell me which is which. And one of these is literally millimeters across, and the other is billions of miles. So, it turns out the one on the left is a confocal microscopy image of a human cornea, and, on the right, we have a radio telescope image of the star forming region NGC-1333. Now, aside from the fact that these images look similar, and doctors trying to find like a tumor on a patient's brain, or a young star forming is similar, the way the data comes from the machine or the telescope, is remarkably similar. So, here's a MRI scanner, and if you've never seen the raw data of a patient's brain, this is what it looks like. And when the MRI scanner is acquiring the data it goes in slices, so you can see the patient's nose, their eyes, it kinda progresses towards the middle of the head; you can start to see the cortex, and it steps through to the back of the brain. Now, believe it or not, telescopes, and particularly radio telescopes, operate in a similar manner. If we were to look at the raw data from these telescopes, we're gonna look at a nebula called M16, and we start with this radio telescope at the front of the nebula, stepping back towards the middle of the nebula, just like the middle of the patient's brain, and all those bright regions are where young stars are forming, all the way to the back of the nebula, just like the back of the patient's head. Now, although the doctors are able to then take this data and look at the 3D, and do surgical planning. This is like cutting-edge, just about as good as what you get with any astronomer and this is what they have to look at, to understand the 3D structure and velocity's momentum in our universe. But we can do better. So, you might recognize this nebula, more like this: the famous Hubble image of the Pillars of Creation or the Eagle nebula. And, I'm gonna fade this out onto a radio image, it's a false color in the background, and fade away the Hubble's image you're used to. But, we don't need to just look at this in 3D, we look at it in 2D, and here I'm using a radiology tool kit called OsiriX. When I showed this to astronomer Marc Pound, whose data this is, he was amazed, because he had been trying so hard to study the impact of a young group of stars, and he had this theory that there's wind crashing and tossing the pillars over, and it took him months to prove this with conventional visualization. But, in one shot, you can see the shock wave of wind blasting through across to the left hand side of the screen. Now, I don't think myself or any of my collaborators would've anticipated how far this has gone, and by sharing the medical technology of astronomy and astronomy with medical, we've been able to find new stars, and supernova remnants, and revolutionize how you do heart diagnostics, and look at data for different patients and organize it in data mind net. Now, I don't have time to show you all these great projects. But I'll show you one of them. This is a collaboration I've been working on, called the multiscale hemodynamics project, working with doctors at Brigham & Women's hospital. Now, what this represents is a novel way of doing heart disease diagnostics. And, instead of the conventional invasive angiography, this is just a CT scan, and what you see here are the coronary arteries. So, you have your heart and the arteries wrap around the outside, these are the arteries you worry about getting blocked, and giving you a heart attack, and killing you. So, it's really important that we look at them. Now, this is a CT scan of a patient with a blood flow simulation, that's the coloring up there; that simulation was originally developed for studying the structure of DNA, and the visualization was done with a tool kit called Visit, originally developed for physics simulations, interdisciplinary. Now, my assignment was to try and come up with a new way of looking at this, to make it optimal for the doctors and hospital; how can we make it the most efficient for them for a diagnosis. And I came up with this image. It is 2D, I took the whole artery and collapsed everything into a 2D plane. I got some very quizzical looks when I showed this to the doctors originally, but I was inspired to do this representation from my astronomy work, where we've been using these tree diagrams along the bottom to understand the structure of Nebulae. Well, we were inspired in that work from the bio informatics and genome community, where they use this tree diagrams to understand their gene expression data. They were inspired by the evolutionary biologists, who use these tree diagrams to understand how species evolve, how they're related, the first of which was drawn by Sir Charles Darwin, and here's an example from his 'Origin of the Species'. So, straight from Darwin through biology, physics, astronomy, back to medical imaging. Interdisciplinary. One may say, well, is this 2D representation better? A study of Harvard Medical School aimed to answer that one question. And it turns out if you present the image on the left, to a doctor, on average they find about 39% of the high risk regions that could explode or block your heart and kill you. On the right, we can do a little better, and they're able to find 62% of these high-risk dangerous regions. But we can do even better, simply by changing the colors. So, the rainbow color map is the sin most doctors and astronomers and physicists are guilty of using, and it doesn't focus on the best qualities of your visual system. Human system can see brightness variation, contrast, not really good at that whole red-green-yellow-blue thing. But now, if you look at shades of red and highlight the regions that are most diseased with dark red, now doctors can find 91% of the high-risk regions, simply by changing the colors. (Applause) And I would never known the importance of color, if it was not for my computer science and visualization collaborators showing this to me; so, again, interdisciplinary collaboration. Now, how do you even get a collaboration like this? In the case of astronomical medicine, it started with a Harvard Astronomy professor, Alyssa Goodman, serendipitously meeting a computer scientist and imaging specialist from Brigham and Woman’s Hospital, and their recruitment of a very adventurous, open-minded, young students. And from there, it has exploded and we've pulled in cardiologists, and computer scientists, and radiologists, and astronomers, physicists, chemists, computational physicists, I mean, we've brought so many people together, and it's been enlightening to share domains and information across borders. And we're still going and although most of the people up on the screen are from Harvard and Harvard Med, now we cross different institutions and continents to work together. And all I can say is, it's been wonderful, we're continuing to make new discoveries. And I just urge you, attend conferences not in your own domain, read books and journals not in your own discipline, watch TED talks and come to events like this, and say hi to the neighbor sitting next to you, 'cos you really never know where your next great idea is gonna come from. Thank you. (Applause)