Transcript: Sean Doyle, President and Principle Architect, Amicas, Inc., 11/15/98

Video tape 1 of 2

C: I guess we can start with dinosaurs, I mean, we are in the midst of one of the greatest dinosaur find in history.

S: true. Very true. I haven't read the paper yet because it's going to be out in science next week. But that's very exciting news. It's an interesting find because the jaws are clearly, they seem at this point, clearly adapted for catching things like crocodiles or for aquatic activity, but I don't know if enough of the the skeleton has been found for them to know if it had, like broader feet for walking on muddy banks. The dinosaurs, there aren't any good examples of them here, they are like the (garble) or therapod? types, but a lot of them have ossified tendons that actually make the tail quite rigid, will they run at very high speeds to the tale doesn't unduly back and forth they lose a lot of energy if it did, but if its foam like a crocodile then you'd want them to be able to oscillator like a fish. So there are all these questions but I don't really know the answers to them, but maybe after the paper comes out, more light will be shed. There's clearly, there's this whole other creature that was never known before, and now, or very little is known about it. Some of the type specimens, I think were found in Germany, but then the collections were destroyed in bombing during World War II.

C: did they know to go to Nigeria to look for it?

S: there have been fossils that been found in Africa for a while, I don't know much about the geology of Africa, I assume since it's a big place is lots of different types of geographic stuff around. But they're haven't been that many discoveries that I know of that came from Africa, there have been a few from Madagascar which is sort of in the neighborhood...

C: geologically speaking...

S: geologically speaking, right, like early bird fossils there seem to be a lot from there. I really don't know that much about the geology or paleology of Africa. I was sooner this point that Nigeria or Niger would be in, I mean there's been so much oil exploration, right they probably have all the different types of sediments pegged, in the new just look for things in the right horizon. And I'm sure there's indigenous African scientist wanna publish and be like people in the West and there probably looking at their own stuff.

C: Yeah I wonder if there's any overlap with the endangered species treaties, or something, there's always political conflicts they're about self-determination, control over trade...

S: I'm sure there are, right now in China actually, I've heard, I'm not sure this is true, it is executable by death to be exporting things like fossil eggs. Because they consider it part of their national treasure. You know I only have the internet for these things [laughter], it's hard to know what's true and what's not true.

C: that's a nice way to begin — " I only have the internet So..."

S: yes, on the internet truth is not just relative it's obsolete. [Laughter]

C: well that I wanna segue, I wanna try my interviewing skills, and segue to, at one point when we had been talking about dinosaurs, you had mentioned a project that you want to do, that had to do with imaging fossils that you would do after this project, what was that?

S: well I haven't thought very much about it and I was too busy this fall to go to be society of Herpo-Paleontology, but they had a special symposium there on in computer representations of fossils, it wasn't clear to me what the scope of that was, it was probably wanna these things where they send out requests, and it takes form when it happens. But I think Ralph Chapman is there from the Smithsonian one of the people who has done a lot of interesting computer work with morphological analysis of fossils. So I'm assuming the proceedings of that will just get published and i'll get them in the mail when they come out.

C: so this would be more like image processing stuff on fossil representations...

S: well there are a few different areas, there's this notion, certainly not original with me that a lot of the interesting stuff in fossils gets destroyed when you take them out of the matrix of the rock, there might be some soft skin impressions or things like that that are really almost gone, they're not really there, but you might be able to find them with advanced imaging stuff. In the latest scientific American, which I was just showing Adrian, there these tomographic projections of mosquito knees, that show the sort of resolution that you can get from these different techniques. But the thing about mosquito knee is that it is very tiny very thin, the noise that you have from looking through a fossil that has rock all these things embedded in it, is very difficult, but computers are getting better a lot of the stuff the computations you could actually doing parallel. So its maybe not insurmountable, clearly not doable in the next year or two. But can you figure out what's there, and you know the might be all sorts of valuable information not about the specimens, about the fossil itself, but you know, about how it got buried and stuff like that, you might be able to pick up some very subtle cues from looking at the layering, the different densities, the different minerals, the different places there. You might figure out how many seasons the bones were exposed to air before they got buried or something I don't know what you'd find. So there's that level of using the computer as a tool for scientific analysis. But then one of the things that you can do, like once you get these things, you can get tomographic projections of these things or you can build models of it, you know wavelet is one of the techniques, you could use all these techniques for building models of, basically polygon approximations of the fossils or the organisms. Obviously this will work very very well within six since there aren't that many insects systematists? people that study systematics in the world it sort of a dying specialty but you know if you could digitize enough of these creatures and, put the representation in a way that you could search by shape, and find analogous structures, in other organisms, that might be a way to help draw relationships between different creatures, you could look at, if you sequence the DNA for different creatures you might find convergent structure in a wing or a leg or something like that, long after these two lineages divided, and that might give you some information about what sort of things were likely to arise, and how the mechanism of DNA expressing itself can be constrained and not constrained, because you can build these better models of the shapes of things then you can now, you can look at, you know if you digitize a whole range of species with this technique, you wouldn't even need, you don't need to kill them or take them apart its just fairly low doses of x-rays. You can look at them over time and establish growth patterns and thenif those, if you could specify what those growth mechanisms were mathematically you might be able to look for differences between species or if you found a specimen find out well its not like any of the once I had before but it fits in between here and here so it's the same species or something. With insects its fairly hard, because they change and they molt, they can change fairly rapidly in body plan between things, but I think it's one of the interesting things in science, you got all these techniques, its true in medicine is well, you got all these techniques that can get you more and more data, but the traditional form presenting them is such that people haven't adjusted.

C: so, in a way, it seems like it's not just a change in scope and scale, but a qualitative change in the way the data is represented, so that you can actually do different kinds of experiments.

S: sure, imagine that, for the home market, you could, if you had these polygon representations of dinosaurs, there's no reason you couldn't print them out in three dimensions, and have life size dinosaur kits for cheap. [Laughter] you know, out replicate Barney on the planet, that would be very important thing to do. But also for scientific study, just being able to, you know compare, I mean right now if you want to measure something on a bone, either you have to call up the person who has that specimen, and asked them to measure it, or if you had these representations you could play around with it, and you can probably find patterns that right now you couldn't. But it's not enough just to accumulate the numbers, you need a way that people can digitize lots of stuff quickly, and safely, and the mechanism for sharing it, visualizing the data, and then I guess part of, just because I'm curious about lots of sort of different things, making it accessible, because what do I know about this stuff other than what I can just you know read in the gutter or in science or nature. You know, I'm a dillettante. And so having the idea that you can, you know instead of having to read these long descriptions, you know that there's this you know, off the sqasimosal? bone, there's this sort of process that's unusual, I can say well so I know that's on the head it's on the skull, but what does that look like in other creatures, and since I don't have a command of the literature, and I haven't seen a lot of these specimens, its hard for me to make in the instant association, Oh its like this in Triceratops but not in Proceratops. You know, to be able to navigate back and forth like that, would be very fun for a dillettante like me. It would be of good scientific use as well.

C: it would encourage more dillettantes.

S: yes better living to dillettantism. So, wind I think about what's sort of next, you know applying a lot of things that I've learned to basic science stuff or, you know not really popular rising science, but somehow qualitatively changing the way data is presented so that it makes it easier to understand, easier to see, because I think the vision system has a lot to do with how we understand things, so using pyramid encoding schemes so you can bring the images back and different resolutions, wavelet are very good about encoding edges, so you can do searches by shape, there's a lot of room for sort of interesting data acquisition, and what we used to call in economics, data mining to, where you're looking for patterns, and sometimes that's a good thing to do and sometimes you generate garbage, but that's or judgment comes in.

C: how you think about the relationship then between building those kinds of tools in say commercial realm, and the role of basic research, because Adrian I have had this discussion in another realm, because of the way the kind of research is proposing, and the relationship to mess general, messes up traditional boundaries between MGH, or academic medicine and industry.

S: we're not the only one transgressing that boundary, its like they've melted. And on one level its like a really upsetting thing, because all knowledge is becoming commodified, you know the notion that people could patent things that we've been doing for a while, because when the companies were competing with wrote some patents, wrote a patent, that you know at one reading of it would say they basically owned all of telemedicine in the U.S. right. And it was a lot of that original work, or talking about it, was done at R*star, my old company. This was a person who had left company then founded another company and did this, its just very scary its not necessisarily that your competing on quality its or whatever, its like that someone could own a basic idea, like the idea of routing images between two hospitals depending on qualities of the images. You know why should that's to be patentable? Who should own ideas like that? You know I can understand patent in some processes or even some algorithms, but stuff like that, it just seems like, your just using the stuff. And then with an academic, the boundary with academic stuff, people are often trying to publish things and then start companies. And there is a world of difference between having a good idea and actually being able to implement it well enough that it could be packaged and sold. But there was enough of a gold rush that was going on, at least over the last few years, that people didn't make that distinction.

C: well I tend to be very cynical about the University, probably only because I'm in it, but...

S: what with it being bought by Disney and all...

C: yeah after that, in all honesty there's a great deal of hypocrisy there considering that the University especially a place like MIT that runs itself more and more along the lines of, precisely structural model of a business, but it has tax-exempt status, get all of this funding from government, it really occupies a special position without upholding that. And I think it's because there's a real crisis of legitimacy about who is actually responsible for that, if the governments not responsible for it then how can the University be responsible for it, so I, I'm trying to figure out where in industry, people experience what they're doing as something fundamentally political, as opposed to feeling like they are, feeling like when they are in the University, they somehow have this authenticity, though what they're doing as political because they are in the University, do you understand the difference I'm trying to...?

S: NoI'm not to...

C: yeah it's very, its probably a little bit too convoluted, but the sense I get with a lot of researchers who are in the University is that they can do what would in, what people would do in a business and industry, but that they have some extra authenticity in doing it because they are in the University, the reverse process were someone is an industry and they're doing was essentially research, they have to worry about patent issues profit motives, where do people experience that, or we're do you experience that in this company? Or where you're future projects involving dinosaurs include that?

S: my future projects involving dinosaurs, I can imagine a world in which they would be commercial products, you know, in my imagination right, the things are would work and maybe with other people maybe work on the Smithsonian Harvard or places like that, the only thing to do would be to give them away. You know part of its like this feeling you know that I've got so much emotional energy from all this dinosaur stuff and I want to give some back. And, the sense that you know like right now, I'm trying to play around with wavelets, and understand them better, but want to buy, you know I don't have time to take a course in that, but you can buy Wavepack 4, or something for met lab, and I forget with the cost is but for me to buy as individual would be something like twelve hundred dollars. And, so what do I do, i'm gonna build it myself. And that takes a lot longer and I'll probably end up understanding things a lot better. But there's this real feeling I have, why should this stuff cost so much, you know. There's this free software foundation dictum, you know, that knowledge should be free, and I don't agree with that completely, I think that copyright and ownership have some real value, but I get defensive when I hear that people are patenting like XOR when they're drawing on the screen, or these things that might be foundational patents for one field, but that everyone uses and shares, but just kind of dropout of the math, for how things work.

C: but if you are in a position where someone was using this, and someone goes ahead and patents it, then you have a crisis there, because either the patent works, and the government enforces it, and everyone drops out, or people say "you can't patent that, what I can of a that law..."

S: I think that's what happened with access's telemedicine patent, as far as I can tell everyone just ignored it. You know, a few years ago at RSNA, they sort of announced that they had it, baby was another meeting because it was at SPIE or something like that, and they wouldn't give a copy of it out, they just said that they had patented it. But you can get it online at the IBM WebServer to. People just as far as I can tell ignored it. You know it's up to them to challenge it, and people have been doing telemedicine for years. And if they had a procedure, where they used to do it manually, and now the computer program says "oh, if it's a CT sanded here and if it's an MR send there" you know is that like patentable? Where in the past someone would hit two buttons or dialed two phone numbers. It doesn't hit me as being that, it's something just designed to hurt people or hurt the field, I don't see it as any real contribution to knowledge, the needs to be protected. But I'm trying to figure out how the applies directly to my company now, you know there's definitely benefits we've gotten from using commercial products things from academic institutions like Michigan state University, they've done a DICOM implementation, you know, we give them a cut rate license for Amicas, and we get some source code for some DICOM work that they did.

C: and that seems like a really valuable collaboration, I mean I remember you saying that these people were really committed to doing this thing right.

S: right. And the department its a lot of value for the they get these little demonstrations of stuff that they're doing at RSNA, its recognized as being high-quality effort, I don't know the exact size of the department but, you know, it's not from one of the large prestigious centers, but they do really high-quality work, and that's been recognized for the most part, so there are lots of good psychic benefits that they get there from that. So there's benefits that go in both directions there—from us they get a commercial product, that has a great deal of clinical value for them, in the short run and they don't have to pay fullprice. So, but in trying to understand exactly your question, rather than being about specific deals it's more about ownership of ideas and where the boundaries are, like when do you say, "this is mine, mine, mine" like a three-year-old, and I'm going to protect it with a whole horde of lawyers vs. this is like what the next step of evolution of this technology is a, and like most things on the web you realize, that if you just hold it to youself people are just going to work around you, so what you do is you put back and hope that this stuff grows.

C: well I have a real sense that things are changing radically with the internet, so does Adrian obviously, and I've learned a lot about that just from hearing him talk, speculate. So, I mean my question is I guess a certain point it's really just a philosophical question, it's not a well specified question or well formed question, because it really is about the status and the role of knowledge and advanced society, and what its relationship is to corporate capitalism, nations and nation states and to individuals, where is control going to be elaborated, what forms of democracy are compatible with this kind of knowledge, obviously its too big question to be answered in an interview.

S: well there is, on a microlevel there's this feeling that I have that I learned how to program off the internet, one worked in Washington at the Federal Reserve Board, they had a very fast internet feed there, well at the time it was fast, I have no idea how I'd experience it now. But they switched over to Unix from IBM mainframes, I learned about C programming, I would go out and buy some textbooks, download all of the source code from the free software foundation and just sort of see how you build a high-quality, low-cost application. There were newsgroups that you could join and ask questions on, you know back in those days, when they're really wasn't spam, and people that like created the compiler, or wrote these different environments would be on the news list, on the newsgroup themselves, you could learn a lot just by listening to conversations or asking questions that someone would answer, you ask a sort of elementary question, or something it seemed elementary week before, and now you knew you could answer it, and there was this real excitement about sharing and that we were all of this together right. Sure a very small corner of human knowledge but, if I think about it like an economist, there was no utility in me helping maintain the ignorance of this person who had asked a question, it was just good for everyone. At times it would get taxing, because they were times when I would say how something works and then I would get like 50 questions the next day. About related things sometimes you can answer it sometimes it's just too overwhelming. But there was this notion that here was this very valuable, it turns out to a been quite fragile, mode of sharing of knowledge and information.

C: well you talk about it as if you have the sense that that's not they are anymore.

S: I think that's really disappeared. When I wait out to Java One, the big annual Java conference that Sun has, I met to James Davidson, I think that's his name, one of the people who was the developer of the Java WebServer which we use at Amicas. And I chatted with a very briefly about the product that we use at or happy with an have no complaints about, and I told him that but I also want to thank him, because right now they have a private mailing list, specifically it's not the Usenet thing, they don't want it to become a Usenet thing, you have to join it, there is no cost to join it, you just go to WebSite and you click on a button and it subscribes you, but they can subscribe you if they want, so people are relatively well behaved, but of the groups that I was on on the net at that point, it had the highest quality questions and answers, and they're really was this very serious attempt to build up this community of people who use this tool. Any said "oh thanks that means a lot to me, you know the internet is really gone in the toilet" or maybe you didn't use those words, but instead of being open now we have to build these little closed communities that we don't really announce broadly and those committees for collaboration but we don't, it's like once too many people get involved, or you get someone in their that does something that sours people on the group. I mean I feel I helped destroy one of the first mailing lists, I think it was a mailing list newsgroup for Mathematica when it came out. Because I found lots of bugs in the first version that I just kept posting them and they were very elementary things, but some people didn't like the fact that I wasn't a Ph.D. mathematician but I was saying oh this thing doesn't work: if you do a matrix this way you get this answer which is right to do it this way it's wrong, but then one of the messages that I sent got sent like I don't know, a few thousand times, to everyone on the list, it was nothing that I had done to do is just something went wild. And so a whole bunch of people just unsubscribed from the list, and I don't know if it ever bounced back, you know but after that you would post things and no one would ever answer, it was like the community had been destroyed, and mine was the only message getting re-broadcast, and itwas pretty long and was something about sparse matrix multiplication, so it was really pretty much on topic, it wasn't like an emotional thing it was just, "oh here's this bug" but I think people were just tired of seeing the same bug over and over again. So yet these things are very fragile, you know on the dinosaur list I'm on, there's this guy mickey roe? Who kicks people off without any public fanfare if they violate the rules of the group. The entomology list I'm on the same thing, there was this guy who criticized other people for being, there was this flame war that went back and forth before the moderator came on and said that only scientific entomology was going to be discussed, and that strong language was not allowed to be used because he was Canadian. And that put an end to it. But these are very fragile communities and you need in guardians, it's really up to people that have some passion for it to volunteer, and that's becoming rarer and rarer. Like I don't know how many hours I work a week but it would be very hard for me to moderate a list right now.

C: but do you think the possibility for people to learn things in the way the you learn to program with a way the you learned Paleo-herpetology without participating in that kind of a group?

S: it would be much much harder, because you wouldn't be able to ask questions, or you know I lurk a lot of lists for like the Java WebServer the Java list the Java speech list, there's probably something like 20 list I'm on, and I just sort of scan through the very quickly each day so that I know without ever having experimented that like JDK 1.2 has major performance problems So its like I don't have to really participate to know that there are other people and that just tells me, "oh don't waste my time" So the half-hour or so it takes for me each day to go through this stuff, or maybe or maybe longer dispersed throughout the day can save me a lot of time. There's one issue that I sort of posted on the net and other people posted on the without answer that was the sort of bug, in my view it's a bug that Sun has in its ActiveX containment stuff for it won't load in native methods, things written in C/C++ when it's inside of an ActiveX component. There's a Sun bug parade were you can look at all of the bugs that they currently have an they say that this problem is closed it was closed a year ago and it will be fixed in the future, but nothing about how to work around it, the work around turns out to be something very simple, something I can summarize but what I'll have to do is write a paragraph or two put it on the bug parade and so that people who run into the same problems I do can find out the answer. Try to put in, you know try carefully put in, because the search database to do so large now that you have to put in all the right buzzwords so that when people do searches they find it. And then i'll post to the advanced Java list, because I've gotten so much from them, why not write this tw paragraph thing give them a link to where the bug is and the purported fix one paragraph in general what you have to do. That's a way of giving back, I feel like it really is a community in that way, but its feeling weaker. Now it maybe the case that if I wasn't under such commercial pressure to get things out you know I have more time to feel like I was participating, would play more with the latest stuff, right now were not using all the latest stuff, well were using all the latest stuff that's been released from Sun, but not any of the stuff that's in beta.

C: the experience for me is the, I've never, even though I've been on the internet since the early '90s lets say, saw it go from being fairly small to being extremely huge, my frustration still comes at finding something that I know I want but can't have access to, and that's just gotten worse obviously. But in terms of finding advice of finding answers to questions, to me they're all kinds of techniques that I would love to be able to formalize and teach to people, that's activity I would really enjoy, teaching people how to use this to learn things. So my sense is that really only a question of what kinds of things are not available, and in graduate school principally that's been various kinds free search, done by consulting firms or research groups or whatever and There was a note from one of the lists you sent me. We agreed that was much harder to get these days was data, however construed, but it was easier to find very sophisticated theorizing about that data when it was available. What had happened was a sort of disequilibrium where it was much harder to actually get access to the kind of data that it cost money and time and resources to get. That seems to me to be the difference.

S: Well I have two hopes of that for the future and there very much I think very unrealistic.. One is that there's work done like that my Tanya Sweeny who is now at Carnegie Mellon is for medical records has been working on these various scrubbing or anonymizing programs that can. You can run them over your hospital medical records and depending on the level of anonymity that you want it will remove patient name, patient identification at the first level, but then they can parse enough of the grammar of the sentences to know that if like if its Hillary Clinton you can change the name, but if it says "her husband, the President," right or stuff like that it can try to catch something like that, or if there's a patient record where someone has these three very rare diseases, well you might just throw them out of the sample because someone might know, "Oh that's that one person." And if there's a way in which that kind of thing can be done, then you might be able to start building up these larger databases that can be shared so people wouldn't have to do so much theroizing. It wouldn't just be in medicine, you could have in lots of other fields, umm, small company financial data. If you have, Just think about what a economist would love. How many small start up companies are out there, etc. Maybe if you contributed your data back you could get back these weekly summaries where you could find out if there was some aggregate notion of where companies of your size, when are they going public, and is now a good time? It would really change the information flows, right? But if you had some kind of body that you knew wasn't going to distort the data and that would of course not be the universities, [laughter] simply, to be the keepers of this stufff, you could imagine that you could start building larger and larger databases of things where it would be very difficult for people to reverse engineer these things, without an example of what people are doing and that sort of knowledge would be sharable. So another notion, again not original with me, thatIf you had a way of selectively filtering the information at very low cost to the people who originally had the information, then they could send that off without fear that it was going to be used against them in the future or exclusively to their competitors advantage. And the sum total of knowledge will be higher. The second hope, and I read the Don Quixote last summer so its in that tradition, is that we need better ways of describing data so its easier to find stuff. Like right now that scientific american article about the mosquito's knee that I showed you. Now there's four reconstructed slices, and I immedialtely wanted to see the whole thing. Well I've already been to the cyclotron in France where the pictures were generated and they had some pictures of a spider's fang, but none of a mosquito's knee. Just like three or four pictures here or there. Now I know there must be some data somewhere, but I couldn't find it. Is there any reason for them to hide it away from me? I can't think of any utility theory that would say that it gives them some big advantage. The images in the Scientific American were quite low quality, what would happen if you contrast enhanced then and produced more details, just you know, my basic dilletante curiosity. But that's hidden from me now. Now I could send this guy email, but I have no idea whether he speaks english. I assume because he's Eurpoean that he does and because I'm an American snob who cant. But that's, I guess there, if there were easy ways to package the information, and right now the only candidate I know of is XML for that, then it would be possible to build better search tools so that when you are searching for Alzheimers the name as opposed to Alzheimer's the disease, how would you find it. Or if you had people jamming the net (and Jack knows more about this than I do but apparently the church of scientology encourages all of its members to create pages that have scietology words that are freqently picked up by the search engines, and they vary them in some ways such that if you are actually loooking for an anti-scientologist web site, you can't find it. It's very clever, its like how do you prevent information from being discovered.). If you had, all that XML is a series of conventions, on one level its very very simple, but what's powerful about it is that you could decied with a group of people like Oh when I want to create a scientific paper this is how I describe the contents of it, if its a medical report how do I separate this finding that might be a finding of microsofts [garble] versus a findng of the algorithm that discovered a certain tumor or lesion or something or thinks that it did. That you might have these two competeting notions of evidence. But you can using the dicom structured reporting draft that's available make an XML document from that where you can make that distinction. If you could search through all the ct scans that had gone through MGH that year. and say oh I want all the ones that had this sort of edge on this area of the liver and had this in the report, then show me all the ones that didn't have this, so you could know which ones you missed. I don't see any reason why that, as a very general technique outside of the hospital. Or even between hospitals, you could compare the quality of care, if you had this common way of comparing reports, which right now there isn't, and a common way of encoding images, so that you could search for edges, which there several proposals for right now. You really could build these larger shared models, but the problem is that everyone is gonna view this as their proprietary advantage.

C: right. And for me it returns to a question of standardization, or as you might say standardization all the way down, it really is a question of getting people to agree at some point external to the technology, you have to say everybody's going to use XML, and then you say in XML everyone's going to use these kinds of tags.

S: right, so one thing to do would be to take, I mean I've just done this very informally but I looked through the current definitions, someone put together draft, actually SGML which is one of the precursors of XML, version of the tags, and it was pretty badly designed or leased I couldn't represent the things that I thought the spirit of the report was supposed to represent , so invented my own tags right, so so what I should really do, is hopefully all run into the guy who wrote the draft at RSNA, I can say "oh, when I tried to put into practice here's with a thing fell short," but we also found from one of the vendors that we went and gave a presentation to, that they view this DICOM structured reporting thing as the DICOM committee stepping over the line, who draws these lines is very interesting, who the "they" is, what happened. You note DICOM's been historically for medical imaging, reports are done in HL7 and the HL7 community wants control of that sort of semantic node for the contents of the report. And so he was this guy who is a member of both DICOM and HL 7, who was at Duke which is the center of the HL 7 universe is far as I can tell, and that there was the sort of vague talk about how the specification had gone over the line in terms of being show from one committee to another. Now I'm hoping that both of these groups get flattened like road kill, by the web. But not in the way that, they have a lot to contribute is a lot of valuable intellectual information there, but I don't see any reason for having a new syntax, or having these older syntaxes when there is going to be one that's ubiquitous in the world. So view have IBM Sun Microsoft, pretty much every big vendor right now endorsing XML, or providing free XML tools, like Microsoft office 2000 apparently will save Word documents as XML, will read XML documents as word documents, its pretty much gonna be ubiquitous, so why not try to get their first, and start building some of the tools before the big commercial pressure comes on because in the the time for thoughtful redesign.

C: heavy been keeping up this company Sequoia?

S: I haven't, but I know about them.

C: because they're the ones that I've seen who have done the most with XML, and I haven't looked very far.

S: they our one of the biggest, there's that community in then there's the Corba community, which is sort of a competing notion, not necessarily but in fact they really are,

C: but all of these competing communities could survive if they needed to by wrapping what they've done in another standard, like CORBA being wrapped XML or vice versa to some extent it seems like a moot point.

S: right but in some ways XML's a lot simpler implement than CORBA, it's a lot lighter weights, and even more importantly it works a lot the same way as HTML works, so it's a model that everyone already understands, where the boundaries for gonna be drawn, you have clients you have servers, with CORBA, you got these different objects they can call each other arbitrarily over the net, it's a much more powerful computing notion, it's been something that's been about to hit big for the last few years, I think that the limited exposure, that I've had to it, imported a good portion of Amicas to it, and then the tools were just too unstable, so I just backed it out, so I wasted like over a month on that. I'm sure it's better now and have the feeling that its conceptually a very nice thing, Adrian's point is always, if we're the only people the world that have this version of the CORBA interface, and we shouldn't use it unless it's good for us internally, there's no one they're the even wants to share, its gonna be easier to convince people to share with XML than CORBA, so I don't know if unanswered exactly are question, I think that your right they could be wrapped in one another, and that will happen...

C: my point saying that, is that with respect to the standardization in the statement everybody's gonna be using this, a can happened to a number of ways can happened because the government mandates it in the people say Okay were gonna do it that way, it can happen because the professional associations and standards associations agree to it or it can happen in some sort of free market, network economics way.

S: right and I think the free market ways the one the which were betting on right now, you know if for this structured reporting demo were planning to do at RSNA, to show people the power that this stuff really can hang together to do reports in the world of images, in this very very lightweight way for the first time, , that there might be some free market drive there. It might be that the way the we've define the Texas intelligently done but it would be better if it were endorsed by standards committee but I think the best way do it is just to get it done and send it out.

Video Tape 2.

C: what is your plan for RSNA?

S: a few things, one is just to go out there into the usual panic setting up things, hoping that nothing to damage to the way there, hoping it's not to atrociously cold, because they hold the doors open, they have these huge doors that they keep open so the temperature can sometimes drop below zero, and its just very uncomfortable... then there's no good food around there, so you get this coffee that's labeled Starbucks, but basically tastes like it's out of some urinal. So there's like getting all this stuff together and it always comes together and then after that I think I'm going to make sure that all the boost we have things in, that there stuff is up and running, and so there's that level of the social preparation, and I'm hoping that I'll meet some people out there that I have collaborated with, talk to somebody from Adac face-to-face, to show them some of the latest stuff which may be to bleeding edge for them to use, but is stuff that we would want them to be using clinically in a few months, and the best way to talk about that is to show it and have a conversation in person. Then there's the showing of the stuff that we have, an Adrian can tell you more about this, that I'm able to because I've been spending too much time and getting the actual stuff written, to fill out the vision. But that we really have a very low-cost alternative to the traditional PACS equipment and that are stuff is much more lightweight, and since it's based on Internet standards, its easier for other vendors to link are stuff into the medical records, so it's got more utility for doctors and and for institutions than buying a piece PACS equipment. You know what usually happens is that radiologist and a million dollars and piece of equipment, and now the capital budgets been moved up to the hospital level, to the administration level, so the radiology carbon has to make case for getting this stuff. But others also the sense that people are saying well radiology got this piece of equipment but what value doesn't have for us, if it turns out that they get their reports half an hour slower or an hour slower, no they've spent a million dollars on this big piece of equipment, they view that is a bad by and they lost. Even though the radiologist made needed to keep competitive or to do something else. So were hoping is that are product fits in as either the interface between the PACS equipment and the rest of the enterprise, or as we make it more sophisticated, we have this Tape archive which is to tiny little box, the hardware costs for the small one is I think 7000 dollars, and for bigger ones in my book thirty or forty thousand dollars that this replaces a half-million dollar unit that it qualitatively changes the notion of who has to make the purchase, they can buy one maybe use it for research prep purposes, or other purposes, so this might be changing what the definition of the market is, because most radiologist would like to play on their own, if it's a small outfitted that fits under budget, so there's lowering the cost making a more open so you can play with it, were going to be showing its ActiveX control, that lets you run inside of standard Microsoft products a version of Amicas and probably the biggest market for this in the short run is people the bill hospital information systems in Visual Basic, there's a lot of people that build like their home-grown information systems, like the whole cast system is written in Visual Basic, if they had ActiveX control to get it would field to them but it was a much more easy tied to digest system that having something the web based even though for the inside hasn't changed all it's still using HTTP to go to the web.

C: this because they're both Microsoft or because of something else?

S: this is if you program your job of thing your Java component to be what they called bean, which is basically just that it supports certain interfaces, you can ask the been about what it is, it knows how to draw itself on the screen, there's a tool that Sun provides that lets you take apart and build an ActiveX wrapper around it, so in principle if you wanna like do a slide presentation in this is something we can conceivably do for RSNA, we can construct one by and into show it conceptually, but if your like a radiologist giving some talk, and you have put to images in a slide presentation like people do all the time, and so they keep asking for these tools to say oh I want something so they can cut pace the images at Amicas the them in the right window level of them into PowerPoint, but you can do that now by saving and disk important into the program it's not that hard in operation. But what if you could say here's a steady and interested in and is going to put that study as a component inside of my presentation, you know if it's on a network in the hospital, they just go to the slides they, upon that screen and there's Amicas, right fully functional, and if they set up, we have this notion that we're there were trying to push also for structured reporting where when you're looking at an image in a study, you might into the person in three the report are doing the diagnosis they might say oh here's this lesion on the liver, but you can only see that lesion when you may be image contrasty enough focusing on the right image, so they can capture that state and say here's the window level value, here's the image of looking at say that into tag often inside

Last Modified 11-Sep-99 9:32 PM ckelty@mit.edu

Go Back to the Start