NIH CLINICAL CENTER GRAND ROUNDS
Episode 121609
Time: 57:58
Recorded December 16, 2009

Ninth Annual John Doppman Memorial Lecture for Imaging Sciences
CTA and 3-D Visualization: Its Evolving Role in Oncologic Imaging
Elliot Fishman, MD
Professor of Radiology and Oncology, Johns Hopkins University School of Medicine
Director, Division of Diagnostic Imaging
Director, Division of Abdominal Imaging and Computed Body Tomography
The Johns Hopkins Hospital

ANNOUNCER: Discussing Outstanding Science of the Past, Present and Future - this is NIH Clinical Center Grand Rounds.

(Music establishes, goes under VO)

ANNOUNCER: Greetings and welcome to NIH Clinical Center Grand Rounds, recorded December 16th, 2009 at America's Clinical Research Hospital, the Clinical Center at the National Institutes of Health in Bethesda, Maryland, an agency of the United States Department of Health and Human Services. Today, the Ninth Annual John Doppman Memorial Lecture for Imaging Sciences, featuring Dr. Elliot Fishman, Professor of Radiology and Oncology, Johns Hopkins University School of Medicine, Director of the Division of Diagnostic Imaging
Director, Division of Abdominal Imaging and Computed Body Tomography at the Johns Hopkins Hospital will speak on the topic, "CTA and 3-D Visualization: Its Evolving Role in Oncologic Imaging."

You can see a closed-captioned videocast of this lecture by logging onto http://videocast.nih.gov -- click the "Past Events" link -- or by clicking the "View Videocast" link on the podcast homepage at www.cc.nih.gov/podcast. The NIH CLINICAL CENTER GRAND ROUNDS podcast is a presentation of the NIH Clinical Center, Office of Communications, Patient Recruitment and Public Liaison. For more information about clinical research going on every day at the NIH Clinical Center, log on to http://clinicalcenter.nih.gov.

We take you to the Lipsett Ampitheater at the NIH Clinical Center in Bethesda, Maryland for today's presentation.

BLUEMKE: Well, good morning. I'd like to welcome everyone to the Ninth Annual John Doppman Memorial Lecture for Imaging Sciences, I'm Dr. Bluemke the Director of Radiology and Oncology at NIH. This is represents a memory are a respected colleague and generous mentor. We are very pleased that Mrs. Doppman is here, as she is each year for this lecture and I'd like to introduce Dr. Chang who will mention a few words about Dr. Doppman and this memorial lecture.

CHANG: I have the distinct pleasure of presenting you with some background information of Dr. Doppman, for those who don't know or didn't have the good fortune to know him. Those who know him, already knew a lot more and enjoyed his presence. Dr. Doppman was born in spring field, Massachusetts, he was a New Englander, he graduated at the top of his class from holy cross. First out of a class of 400. Not second… first. He worked in the summer selling good humor ice cream bars, I included that because he enjoyed that but he was also a man of good humor. Everybody loved him. He was encouraging to everybody, very enthusiastic about his work in radiology and his enthusiasm became infectious to the whole department. John Doppman considered entering the ministry but we're grad he went into medicine, not that he didn't have a good influence on the department but you'll see from the next slides that me made huge contributions field of medicine. Graduated from Yale Medical School. He was for many years the Chief of Radiology here, transitioning us from plain radiography all the way to CAT scan and MRI. And in recognition of his accomplishments that were recognized internationally, he would you know the SCVITR gold medal, in 1997, the highest honor you can receive in this society.

Dr. Doppman was a physician first, he studied diseases and that was where his research was, he studied diseases where the treatment or diagnosis was at fault or faulty he improved them and although he wrote many articles, his main impact is in patient care because his outstanding insight and advances were immediately applicable to diseases which had either poor diagnosis or poor treatments. In 1968, he performed the first spinal ambluealization, but in order to do this, he had to study the norm-anatomy as well because there was very little known about normal spinal angiography until he decided to take an interest in this field. He involved several embalmization agents which are used today. One was ivalon, which was one of the first non-restored agents and second was he developed a way to make crazy glue, more opaque so he could see where it was injected. As I mentioned he studied spinal air toriol anatomy and wrote the key textbook in 1959, he introduced the concept of mouse inflammations and if you could treat it successfully would obliterate the AVM.

His other main area of work was in endocrine tumor localization that is they are small tumors they're hard to detect by any imaging technique, even today it's difficult to find many of these tumor but by combining the use of catheters and sampling, he was able to develop ways to find this and help patients avoid fruitless surgery. One much them was in sinus sampling which he promoted which allows people to sample blood from the pituitary gland to determine whether it was access production of hormone from the pituitary gland and he was also developed ways of localizing small eyelet small cells of the pancreas as and they produce hormone hormones 1 of them being insulin and when these are in excess, they cause problems for the patients and these went through procedures to localize these tumors and they are used today worldwide. Adrenal vein sampling and parathyroid localization and parathyroid adenoma, and you could use these concentrations to cause apopitosis and I guess cure the ablated adenoma was that producing too much hormone. And finally we were glad to present another super star here, Dr. Elliott Fishman who was probably the most well-known person in CT imaging who presented the ninth memorial John Doppman Memorial lecture.

BLUEMKE: It is my pleasure to say a few words about Dr. Fishman before we get going here and he did have the pleasure of knowing dr. Doppman, as well and I think his lecture will be in the true spirit of the a lot of the issues that Dr. Chang mentioned as far as Dr. Doppman's level of enthusiasm and expertise. Dr. Fishman is professor of Radiology and Oncology at Johns Hopkins in Baltimore, and he is a Director of the Division of Diagnostic Radiology and Abdominal Imaging and CT there. Dr. Fishman received his M.D. From the University of Maryland and did his residency at radiology at Sinai Hospital in Baltimore. After that he completed his fellowship training in the early days of computed tomography at Johns Hopkins and this was in the same program who also trained another very famous radiologist who we know very well, that's Dr. Elias Zerhouni trained in the same program and just about the same time with Dr. Fishman. Dr. Fishman is really probably the worldwide leader at this point in developing new techniques and technologies in ct, in visualization and post processing methods. He's pioneered 3D imaging in work and his research lab, starting from the days where Pixar is a small company, started to get involved in advance visualization in conjunction with Lucas Films and these were three dimensional images that were starting to be applied 15 years ago or so in CT, in terms of advance visualization of the detail that was present on CT scanning.

Dr. Fishman is an extremely well published author of over a thousand peer reviewed publications. He has authored 8 textbooks and mentors dozens of radiologists, there are at least 20 or more radiologists in terms of fellows and residents per year who come under his direction in the programs at Johns Hopkins. He's a member of multiple societies and fellow of American College of Radiology Society of Body Tomography and MRI. And besides being well known through all the publications and traditional sense, he has the single most popular web site in terms of medical radiology called “CTisus” with over 50,000 users and there's countless numbers of residents, who learn from this site throughout the country and throughout the world in terms of both radiologist and trainees. Dr. Fishman was named America's top doctors list since 2001, outstanding educator, as well as outstanding researcher by popular vote of his peers antminni.com and in 2007 named as the nation's top radiologist. This past year at the RSNA, which is the world's largest medical meeting in Chicago, he was named the outstanding educator of the year. And in addition to those, all of those accolades, his direct impact on my own career was quite immediate as a colleague and mentor and we're very pleased to have him here, one of my happiest days to have Elliott here and he's presenting on CTA and 3D visualization and oncologic imaging. Elliott?

[applause]

FISHMAN: Thanks very much, Dave, it's a real honor to be here and when Dave asked me to give this talk, it was an honor to invite me, now that he's your director and I worked with him when he was a res debt no more than 5 or 6 years ago, it seems and also for the Doppman lecture and I remember Dr. Doppman, and I remember when I tried to publish arms when he was the co editor of that journal he would have to deal with him and pray that he would accept your arms and I guess he accepted a few of them that probably shouldn't have been accepted but he was always very nice, it's truly an honor to be able to give a talk in his honor today.

So per rules I'll give you my disclosure, I'm a consultant and we get research support and I am looking at the graphics so let's go into the story of what we'll speak about today. Oncology in terms of imaging, regardless whether you're with the NIH or Hopkins or any place in between, about 60 or 70% of all the CT we do, relates to oncologic imaging and the goals of oncology and CT have never changed since the late 70s or early 80s detecting disease, defining extent and then of course patient management and each of those steps which i'll look at today, we can do so much better now of course with the changes in technology and also so much of what we look at and things we really do speak about is in radiology we create images so in many senses, what we do is we create assets, we create information.

So our goal is changing data into knowledge and changing that knowledge in a way that it manages patients better so we always speak about the concept of acquiring the right data, doing it correctly, the rote protocols, getting the information to the right person, the referring doc on the right device these days with the new technologies that might be your iPhone at the point of care. So trying to do everything in a very, very fast effective manner is what we're looking at. When you look at technology whether it's hardware or software, obviously the developments at companies like Cisco and Intel and the like, they've really driven everything we can do in terms of imaging but it's more than hard so mere than the softer capability in terms of how we use things. It's really the end goal which as Dave mentioned with the Dr. Chang mentioned with Dr. Doppman's real goal was optimizing patient care and improving outcomes. So that's our goal. So everything I'll speak about goes to that one direction.

If you look at CT, it's amazing how rapidly CT has changed. Beginning change was slow, but the last few years it seems like every year there's new and better scanners and that will continue over the next number of years. The real point, strategic inflection point was at 64 slice CT when it came along, basically everything became isotropic, everything became volumes, and also everything became massive data sets, we're no longer looking at slices. I think CT was a modality that initially focused on slices whether it was 30 slices or a hundred slices, now we begin talking about thousands of slices, massive amounts of gigabyte data sets and there's no doubt when you look at those type of data sets, the ability to understand, interpret and use that information is not something that can be done on slices. If your packed system was fast enough and your risks weren't developing carpal tunnel, you're still not going to get the information from looking at slices. If you think about it, if you look at these 6 examples, each of these cases now, with the newest scanners like you have here can be obtained in the time it takes to simply rotate the data sets. That's a total acquisition time. So doing CT now, these abdominal scans issues in under a half a second. So you think about where you're going, very, very, fast acquisition, the protocols for doing CT are very simple, and those—many of you don't do CT but for everything I'll show you today, the preps are easy, most of the time we use water as a contrast agent. We give intra venous contrast and the volumes of contrast have come down over time. Old days maybe 150 was a common amount of contrast, now its more 80 to 120 or 80 to 100 CCs is more common. We inject now faster than ever. Lower contrast volumes a. increase cost to the system but there's potential issue with renal toxicity, if you stay under 100 CCs, issues do not occur. We do different acquisitions, there are many different phases we can do if in terms of CT and angiography, we limited phases because we want to minimize the radiation dose to the patient. This is a week, maybe not a great week to speak about ct, all these articles coming out about the radiation dose, sort of a rehash of things that have been there before but it does make the point and if you look going forward, the radiation dose that we give patients now, compared to the article that was published on Monday which is roughly 2 year old data is 30% less in general and then if you look at the cardiac cases they showed, we do about 95% less.

So in the sense, the techniques we speak about are being driven with lower and lower radiation dose but we still try to minimize what we do and we don't do everything we can do, but we do what's indeed necessary. We look at how we deliver the contrast there are a number of different techniques to use, often fixed delays which are the simplest process or something we routinely use, but with these faster scanners where the error time, swree little room for error, we do things more like triggers, we're able to get a specific point and then begin to scan and the reason that becomes important, if you look at this schematic, here's a typical 64 slice scanner if I was doing your pan kri as or liver, you would see about 23 centimeters, it takes me about 10 second which is pretty fast but if go to the scanners, it's taking you under half a second. So it takes under half a second, I like to make the point that when you press the button to start, you're also pressing the button to stop. So, if the contrast is not there, it ain't going to get there. There's not enough time.

So we typically do things like this now, where we pick a point in the artery, contrast is coming in, low dose scans, still a contrast hits a certain level and because the cancer is so fast, we trippinger at 250 house field units and we trigger and then have very good pacification and everything is perfectly timed. So the variation that happens between with individuals with older patients, parents with variable cardiac output, all those problems tend to disappear. To get very, very, high quality data sets, so when you take those individual slices, and you try and look at them in a volume and here's just a sagetial view, celiac SMA, you can see the quality of the data set is perfect because the density of contrast at the top and the bottom of the aorta is exactly the same because the scan times were about .5 seconds and you're able to basically raise motion. We also talk about everything we speak about in terms of imaging as beyond the axial display and for those that don't do a lot of imaging, let me show you a few examples why, give you the principals why, everything we look at is either in volume rendering or maximum density projection or MPR, other display where you're not looking at slicings.

If I took this patient who has crohn's disease and I'll show you a non-oncology application for this example. Here are slices of the small bowel. And here's another image of the small bowel and you see this lupus disease is thickened, it's abnormal. Now, you can make a diagnosis but what exactly is going on? What's the extent of disease? Well instead of looking at that 1 slices slice, if you took 1500 slices, you can create a crohnal perspective and here it is here and there it is there and you can see it butter. You can see the length. You're not looking at just a single loop, you're looking at this volume information but again a crohnal view is essentially a slice. You can look at a slab and scroll through the crohnal views but you have to then in a sense, build up the information in your mind or you can take the same data set and look at it in 3d so now in this image I'm accentuating the ability to look at the vessels. All of these vessels were images I showed you a moment ago, but you didn't appreciate them. This is the vases rectus, these are all under a millimeter in size. Special resolution on CT is typically is about .3 to .35, and here's volume disease, where you see the hyperemia and you really can understand the increased flow, that this patient is now with Crohn's disease but active disease is patient being treated aggressively and now instead of those slices, you have the volume and the volume that's substantially more information. Another example, if you look at this patient with the pan kriatic mass, potentially serious tumor, if you go from the axials to the crohnals and if you go from the crohnals to the 3d volume rendering, can you see how the ability to look inside the lesion, the ability for us to know benign cystic pancreatic tumor by looking at the acceptitations, look at the vascular map which becomes critical determining, if you were going to operate on this patient, what they can be operated on or not and taking the volume and then going into hip. To the ability to look at that as a volume changes the information.

One more example. A patient here who had a stenosis of an SMA and a stint was placed and you could see this stint around the vessel but it's hard to appreciate if it the stint's in the right place, is it working, is it patent and the silliac and SMA with the stint in place, you can see the stint nicely, but you can't tell what's going inside the stint so what you do, basically the pink line here we put points here and here and track through the stint and now we open up the vessel, herate the vessel, there's the stint, things are positioned perfectly rings everything looks good. So the ability to get information and look at information is the key to everything we do in terms of all of the paisk imaging and when you start looking at the literature, it's not just visualization for the sake of saying this looks prettier, it's a nice picture which is often the case as well, but if you look at the articles published across a range of applications, it's somewhere between a 20 and 30% difference in the accuracy of looking at volume images verses looking at axial images and that's across a range of processes. So let me show you some examples and I want to focus on is the GI tract and focus mainly on pancreas and live because of the liver because of the responsive time. When you start looking at the pancreas, what do we do in oncology? Early detection of disease is the goal with the goal of surgical intervention in a timely fashion. Many patients probably the majority probably close to 80% at time of presentation are not resectable so what we want to do is determine who is resectable verses who isn't resectable and do the rapid triage, there's no sense operating on someone who's not going to benefit from surgery and then radiation, chemotherapy will become critical at that point and you can see the accuracy of ct, here's 1 article where CT was done with multiphase acquisition to determine whether patients were or were not resectable, again, the goal for us would be we would never want to put a patient in the position of someone really having been resectable and not get operated on so if we're going to err we're going to err on the undercalling instead of overcalling but in this article, the value of CT was 100% and when you compared the path was only 83%, but you can see the very high accuracy and it's not just that, that study, there have been a number ever others like this one by zap bony, again, very, very, high acrase and good reproducibility, and the ability to define and separate patients becomes very critical.

You also then create the capability of going beyond simply looking at the axials, and with the 3D, the ability to determine resectability as articles from Hopkins by Mike House for example, which showed the predictability was about 95%, so we really can be very accurate in terms of those processes, as long as we do the right acquisitions. If you only did the axial imaging, it would be impal to do. So, in saying that, it also has changed how we think about pancreatic cancer, so for example, I mention that we typically put things in resectablable, unresectable categories so now the pancreatic cancer have changed a bit when we seek about resectable and burpedder line resectable. Where some patients obviously will any right to surgery and others will get combination chemee therapy and radiation therapy and then go to surgery, but you recognize the ability to create this border line category and you see some of the comments of what is border line resectable in case of a short segment of the hepatic artery without involvement, above of the SMA but less than 180 and short segment inclusion of SMVor portal vain or confluence. Well, each of those 3 things if you go back 5 years or so would have made the patient unresectable and today we're able to resect and the outcomes are indeed very good. And it's really this, this article mentions, it's the advances in imaging and of course these advances in surgery combine that are allowing you to really help this group of patients that we see many patients are border line resectable and we'll cue in and get chemo radiation and then 6 to 12 weeks later we can get surgery and in some cases, the patients are unresectable and at that point will come back and have negative surgical margins, so again this whole concept of border line resectable is really based on our imaging and it's also shown in these ARLs that it's very clear that you need the expertise in doing that multidisciplinary conferences, high volume pancreatic cancer centers make a big difference and there was a an experience from Hopkins as well that when we look at the preliminary diagnosis and preliminary staging where the patients came to the hospital, and then our redoing the patients and looking at staging of that point, that about 19% of patients had a change in their clinical stage and in fact from the whole conference about a quarter of patients had a change in their recommended diagnosis and recommended management.

So again imaging really does play a major role in this and most of the changes were based on imaging or imaging suggesting different possibilities including the patient might not have pancreatic cancer. Now if you look at pancreatic cancer detection, in many ways we always think about looking for masses, if you go back to the old literature, we talk about pancreatic masses. Pancreatic masses, we still see the masses but the key is picking up lesions when they're very small, often when there's not really a mass present because we're not just looking for size issues these days with high resolution CT was lost the images, we have available, we can look at profusion changes, enhancement changes. We look at duct transition and the like and we're able to really define based on tumor appearance or changes in pancreatic appearance, early tumors so for example, if you look at this case, first thing we notice is that the patient's duct is mildly dilated and that's always something that you have to try to figure out why that's the case but if you look at the mid-body of the pancreas, you see this textural change here, there's not really a mass here per se, we're not changing the borders of the gland but the texture is abnormal and when you see the texture change like that, that's a tumor sitting right here which is about a son o meter or so and it's just a few more images when we take it into volume rendering, you can accentuate the textural differences so we spend effort now not just looking for tumor mapping or this example where patient is in vague abdominal pain but you notice the duct is dilated and when you start looking at the images in different perspectives, trying to figure out why is this dilated, you recognize when you get the images just right, you see this transition point right here and here's the patient's tumor, you see the textural change becomes very, very, important.

So we're able, by looking for these changes. Another example here, dilated pancreatic duct, abrupt cut off, that tells you there has to be a lesion there, but these are commonly the patients we pick up the lesions when they're small enough that these patients in this example have no spread of disease and potentially are curable in that regard, so transitions become important whether it's pancreatic duct or common duct. So for example in this case, we're looking at common duct, again I'm not showing axial images because if you want to follow the duct downward, here's common duct and you see the transition in terms of caliber, you get a patient water, and here's the textural change in the pancreatic head, that's about 2 son o meters worth of tumor without any invasion, here's portal vain, which is good, we get many or images of that.

SMA will be separate, again, very easy to visualize and fortunately in this case, you also see this lesion here, patient had a liver med, so even patients who don't have local spreads where everything looks good can still have metastasis so you have to be careful in that regard.

Another example here in terms of ductal, here's the cut off, see the subtle changes in textature? There's a tumor there, there's no vascular spread, this is the splenectomy, this patient has a great chance of getting post op chemo therapy but the patient has a terrific chance of beating pancreatic cancer pancreatic cancer. So we detect the presence of mass, and we determine whether they're resectable by looking at many things including the vessels. The main reason that patients are not resectable relates to vascular involvement on the both the arterial and venus side and we do this map nothing great detail because we don't worry about looking at slices, and look at volumes, variations in anatomy in this case, the patient comments to silliac and SMA, we can create nice 3 dimensional maps of all the vessels and then we'll go with the vessel, vessel analysis, so for example in this patient, you can see here's the patients sma, but you can see the vessel becomes narrowed, the soft tissue around the vessel here's a few more images of that patient.

That's a classic example, the vessels patent but once it's infiltrated and surrounded 360 degrees, that patient is unresectable. The patient didn't have a large mass but it's infiltration that makes the patient unreceltable, here's a second example, very similar appearance again, just showing you very subtle that that patient's tumors infiltrating that patient would not be resectable. So this example, here's another patient with tumor, the question is what about extent and can you see here, here's example, SMA looks good, although the tumor above it but the silliac encased and again this patient would be unresectable. You can look also on the venus side of things, so for example here we're looking at the patient, the SMA looks good, the silliac looks good, the hepatic artery looks good, this GDA is encased but the GDA rifles with those procedures anyway and the GTA is not important so from the arterial side you would say this looks pretty good and here's the patient's mass, pass is about 3 meters, not too large but unfortunately when you go to the venous side, here's what happens. Here's the SMV, you see this gap here, the tumor has occluded the vessel.

Can you see collaterals here, can you see increased flow into the mesentery so this patient is resectable. We mentioned short segment involvement. You can do surgery but once there's encasement, inclusion, once there's collateralization it's not going to happen. So again, we're looking on both sides of the fence. Another example here, on the arterial side, this patient has a stint in place, can you see nicely, the silliacs, splenic hepatic artery, branch vessels off the SMA, renal arteries. So far so good. Here's the patient's pancreatic mass but as you look more carefully to the venus side of things, that pass unfortunately involves the portal vain confluence, it's narrowed but it'sinarred a bit more than what 1 would like and it's also narrowed by the smv and this type of patient will get chemo therapy and radiation therapy with the hope of potentially making this patient resectable though in this case, it was not helpful.

So again, how we look at things are very de-opinioned spent how accurate we are dependent on techniques we use and whether it spread beyond the pancreas and in this case, you see the renal artery and vain are involved and if you look a bit more carefully, the mass infiltrates and it was obstructing the patient's duodenum here, so you can see that as l. So again, the extent of spread, planning what need tubes done, overall shown on the volume, in terms ever processing time, the question always is, how much time does it take to do this, who does the imaging, we asked the radiologist through the post processing, typically it takes less than 5 minutes to do the process and it's not just an aden o carcinoma, Dr. Chang mentioned 1 of the Doppman's areas was endocrine tumors and eyelet areas and when you go back to the 80s and 90s, the accuracy of CT detection was 30 or 35%. These days its 95%, here's the reason why, here a 1 centimeter eyelet cell tumor, can you see this bulging off this lesion here. On those images but when you take them to 3d it becomes much more obvious, there's the vascular lesion and there it is again here. So it's not just aden o carcinomas, it's it is range of pancreatic tumors going from larger vascular lesions, a large eyelet cell, vascular invasion, liver metastasis, the smallest 1s and we're now in the 90% range. And renal cell carcinoma 10 to 15 years later can present, and hyper vascular and look identical to eyelet cell tumors and no again no real mass effect but it's the vascularity, and only by scanning at about 25 to 30 seconds post injection, the arterial base images can you detect the lesions if you go back and scan it a minute, pancreas looks absolutely normal and again a lot of what we do in terms of images and the accuracy is dependent on the protocol and unless you're scanning early and fast enough, things are not going to work out.

So of course as I mention with the fast scanners it's just ideal. The other thing these days we look at are cystic pancreatic lesions and I'll show you 1 example. We wrote a paper from Hopkins at about 3% of patients have incidental cystic pancreatic lesions. No history of pancreatic dancer, no family history, we put all the patients aside. Hundred parents come off the street for an abdominal CT, three will have pancreatic lesions and the question of course is what do do you with these lesions, there's about 10% chance of displastic changes leading to carcinoma, and typically the rule these days is cystic lesions, water density and so cceptitations no nodularity, no enhancement, no clinical symptoms related. Simply follow the patient 6 months and then a year, se went sequentially but we're able to look at these lesions so in this case for example, when you look at the lesion, you see the dilated pancreatic duct, caused by the cystic lesion, most of the benign lesions are what's called side branch IPM and intra ductal pathular, and you this 1 you look at and you see the detail of this nodularity and I'll look at this image in multiple plains and when you look very carefully beside these acceptitations, these are supet areas of enhancement and this lesion's a little over 3 sonometers and it's not a lesion can you leave alone and this patient went to surgery and at surgery they show, the path, this was one of the areas that concern me, this little nodule here, this was an early malignancy. This patient had the whipples procedure, the this patient should be cured. Again, it's very early imaging, but it's very subtle findings but we can see these in 3D volumes.

Okay, what else? We look the liver. The concept of liver analysis and liver detection of disease, whether it's primary metastatic is based on a number of things including how the lesion behaves with contrast, vascularity and the like. And in the past, we also had the issue we couldn't scan fast enough and we think about contrast being this pink curve, if we inject contrast, at 5 CCs a second, the very fast upslope and very fast down slope. This area here is arterial phase and late phase, arteriole phase is critical for many things, if you look at endocrine tumor, metastatic deliver, it may show only arteriole phase. If you scan here at positive, if you scan here it's negative. Hepa tome as will only be seen in arterial phase imaging and often they're very subtle at that point and wire doing Hep B and Hep-C and that becomes critical. How can that help in that regard? So when you look at the liver. That's same concepts hold true. We can look at the liver as a volume but within that volume, we can determine what we want to see so we want to see the hepatic artery nicely shown. Or so another case of a common trunk or scanning a bit later where the patient has cirrhosis and these 2 little structures here are recatalyzed umbilical veins and in 3d, you see the vessel sitting along the surface of the liver as it's reconstituted there. And there. Or this patient was cirrhosis, we have a combination of hyper proanemia so you have this wet bowel pattern and you increase flow, the mesoterric vessels, nicely shown here in imaging and here in volume rendering, so it's capabilitys, arteriole side or venus side, have you and again this is just from a hundred CCs of contrast, peripheral injection, at about 4 to 5 CCs a second.

And so, when you have occlusion of vessels, here you see, not total but 70% thrombosis of portal vain and then you see the collateral here, kaffennous transformation which you then can put another perspective, thromboas, again, much easier to see extent of vessel involvement both distance as well as cross sectional volume and then the collaterals and coronal as well as volume rendering in it. So, we have the tools on both the arterial and venus side. So how do we put that in play when we look at patients in terms of masses? Well the simplest thing, 8 10% of patients have cysts, most of the cysts are small, this is a larger cyst but the rules, water density, here it is crohnal, well-defined margins no discernible wall, but it's mass like and you'll see splaying of the vessels. So a critical thing for us is to look at the vascular map. Within the vascular map we get lots of additional information. So here benign lesion, displaying of the vessels. Mass effect, no problem. One of the most common lesions we see that's benigner manche I don't mean as, they can be confusing particularly when they're an oncology patient when we have a lesion that might be vascular and it's a classic pattern. Early phase imaging, they have puddling around the edge, and over time the lesions fill in from peripheral to esebtial pattern. Others become dense, others might have the scar but it's the early appearance that makes the diagnosis. And it's significantly different than metastasis, significantly different than vascular hepatomias in terms of appearance, within the 3d volume it's much easier to appreciate the way the pudding looks, you can see here. You also can see a feeding vessel not uncommonly but the feeding vessel typically is small and there's no irregularity as you see in the neoplastic process and then of course over time as I mentioned the lesion can go in, but it's very, classic puddling, central scar, classic for manche m ang ioma.

Now you say how do I know this is not a metastasis, the neurobedocrib tumor so you see it has an enhancing rim but it doesn't have the puddling we were looking at a moment ago. It doesn't have that puddling but it's enhancing rim but also have you the neovascularrity that surrounds the lesion, so have you the micro invasion adjacent to the lesion and then when have you vessels speeding so have you irregularity so what we're looking at are signatures of lesions. So the ability to say, yes we see a mass, we can do better than ever but seeing a mass doesn't mean a whole lots. A mass can be de9ed it can be malignant. Oncology patients a number of articles show that 1 son o meter lesion without looking at it, is greater than 50% chances it's benign but characteridessation becomes critical. So example, this patient has a large mass in the liver what you reels is this lesion, realize this lesion is vascular, it's only as vascular as the IBC, and it's not as vascular as the aorta. The endocrine tumors are vascular like the aorta, it's homogenous and most primary tomb areas are not and then if you look more carefully, you see there's large vessel, feeding the lesion which is indeed very common and that classic appearance that arterial hyper vascular lesion feeding vessel is really what you are going to see and has another example of a patient with FNH. So we're able to have very characteristic appearances of the lesion and you can see that with time another example, same thing, IBC density, central scar and here's the same lesion over 1 minute and can you see at this point, you really don't recognize there's a lesion here for the most part, you can kind of see a little bit of a shell of a lesion but you can see the ability to identify the lesion here, be specific this, is a leave alone lesion, FNH, common females, it's benign, no malignant potential. It's not a problem, you leave it alone. Oppose this sample here, can you see the vascularity of mets itself looks like the aorta so it's not a problem; you not make that mistake in diagnosis. So signatures become very, very critical.

So when you look at tumors, the same signature becomes critical. If I was asked what the best phase of looking at hepatomia is, it's looking at the arterial phase. If you don't look you'll miss 34%. Arterial phase you see the vascular invasion, the arterial artery involvement nicely and you go from there 30 seconds later, you still see the mass but you can't tell much else. When you look at the latest scanners looking at the liver and these… the first couple cases I show you large tumors to make the point. Yes we see a mass here but as we do 3d imaging and look more carefully ha we're really seeing is neovascularrity, okay, we're looking at vascular maps, so whether you are going to plan surgery or chemo, whatever you want to plan, the neovascularrity becomes critical and if you can see at hard to appreciate in axial imaging there's large tumor and there's vascularity but it's not impressive on these 2 images, right? But if I take that same image, here it is crohnal and then I run it as a mip image, look at the neovascularrity? So all that neovascularrity is there, we just tend not to appreciate it. Here's one more example. Large mass, vascular and when you start going inside the lesion that you're appreciate the neovascularrity. If you take that principle of looking at neovascularrity, on large tumors, you say, not that helpful, I see the tumor but what about smaller tumors but serotic patient, tips catheter, what's this here? This is common angle don't mean as should I biopsy it, the biopsy these patients, there's the lung sitting here, maybe you're not certain what this is. Well if you take that same map and now you put it in 3d, look what happens? You see the vascularity and you see this irregular feeding vessel and you know it's neovascularrity, all the regular, that's a hepa tomia. 100% certain, changes how you manage the patient. Or this patient, patient with hepatitis c, you see changes in the liver. Is this of any significance? You often get ab shunting in these patients is this of any value, well, it's hard to say much about that on the axials or crohnals, but look what happen when is you go in 3d, you see the neovascularrity and you see the irregular vascularity and you can recognize this has to be hepa tomia until proven otherwise. It's the neovascularrity that's only appreciated in the volume that gives you that information. Sometimes it's easy, sometimes, you know the vascularity is so great in the lesion and the feeding vessels are so large that in fact it's easier but you recognize that you are looking at additional signs and in this case, accessibility, but you're looking at the information far more than you can get. We also recognize that the lesions that are missed can be found. This was initially read as negaative. Well there's a little bump here but serotic livers are often bumpy and when you start looking more carefully and you start processing the data, you recognize there's a real blush here and you recognize there's a real feeding vessel here and you buy the new software that comes with the yellow circle that automatically surrounds pathology and it's very, very, easy to see that. Now we've worked on that at Hopkins, Dave and I but we're not releasing it because we both want to keep our jobs so, that software will not happen, but you recognize how easy it is to see. Even if patients who have tumors that are hypovascular, these profusion issues make a big difference. This patient has a lesion here, two lesions have pancreatic cancer, has society, it's not resectable, what are these lesions, this cyst, this tumor, if you look at the venus face, hard to say much, you look at the early phase, this detect here? Well that change mean this is a malignant lesion and that's a simple cyst. So again the profusion changes become critical or this example, this is a patient being followed at Hopkins who had this scan, couple years before I had another scan, looked similar. Patient has hepatic and storage disease and hepatic carcinoma, and they wrote a lot about the disease. They say these patients with develop hepa tome as but follow thanksgiving patient and you say, lit hasn't changed any, just come back in a year but if you took the data set and looked at the 3d map, you see this neovascularrity and with that irregular vascularity, that's not a lesion you leave alone that's a lesion that's going to come out and that came out and to no 1's surprise that was a hepatema, so the ability to detect the lesions, it's a carcinoma patient, you either multiple liver mets but if I showed you the axial images, it was read as negative and in retrospect, it was very, very, hard to see those multiple tiny lesions, in the volume, you see things significantly better.

I'll show you one other example, small bowel to give you a feel. I showed you a Crohn's camp wide receiver, here's a carcinoid tumor, the global visualization taken in this case, roughly 1800 slices, looking at the mesenteric vasculature, see the encasement by the carcinoid tumor of the patient's SMA, looking at the bowel pattern, again, global visualization of the information becomes critical to us and whether it's infiltration by carcinoid or this example. This is the newest scanners. Here's a schemeatic of the artist draws but I like the way he wants to negotiate his salary, show him this image and say look, I do better scanning patients and they're from free. You can see the details you get. The millimeter vessels and this patient when you look carefully had a mesenteric mass, which had been missed initially and it's right here in this tumor and the detail we get with the new scanners, the combination of arterial phase to venus imaging makes all the difference in the world. It allows us to pick up tumors that would have been missed. If you look at small valve tumors, clinical presentation to detection tends to be 6 to 12 months so in that regard, it becomes very important that these new techniques allow us to pick things up much earlier. And this whole concept of imaging of visualize egg, across many areas from renal imaging to gastric images this tumor in the stomach, accurate staging and early detection of tumors, within the volume things are easier to see, and more accurate to see and whether it's adeno carcinoma, or it's something like gastrostromal tumor. It's the same concept, it's a concept of better acquisition and better visualization and again as we say, it's not just simply detecting lesions, but basically getting the information in a format that someone can use it. If you think about the radiologist and our job is detecting disease, defining and staging, we then transfer the information to the referring physician who might be the oncologic surgeon who needs to operate. Having a thousand slice system not going to help anybody and having 2000 slices is not going to help anybody. I remember the old days, many of the physicians like film, the reason people like film if you ask them is because they took the film out of a jack and they hung it up. There was a red circle that surrounded pathology. People like the red crayon. No 1's invented a red crayon for imaging but I think we're getting there and the visualizations allow us. Again, axial imaging alone limits our capabilities to do state-of-the-art imaging, the oncology patient, a volumetric approach is proven and continues to be proven as to way to go in detection and staging of disease, the ability to really rethink how we do things, the delivery of clinical services, the delivery of information is rapidly changing, I think it's truly an inflection point in terms of how we distribute information, I mentioned we showed you this slide before about the right data to the right people on the right device. Now you'll be able to send that information to remote sites whether it's an iPhone, think tablet, anyway you want to send the information whether it's here or around the world. And a place like NIH we have managing patients around the world that opportunity has never been greater to be able to share information. I think in radiology, Dr. Doppman said 30 years ago, opportunity vs. never been greater, the challenges have never been more difficult, but the future has never been blighter. Thank you very much.

[applause]

BLUEMKE: Thank you very much there's time for any questions if there are questions following the talk?

QUESTION: I want to ask you one thing, there's an issue regarding so much information that we already have, it's going up, how is that going to be handled you think in terms of medical treatment decision making really.

FISHMAN: Well, I think I heard a talk by a defense company and they made the point that when you do in defense, you talk about 9/11 or any of these events that go on every day that you're getting satellites and you require incredible information but if you don't use the information, you know, it's worthless and I think it's the same thing in medicine. And they were talking about this transition in medicine that we need to do is we acquire data, the data is not information and so really it's these tools that allow us, you know whether 3D imaging, whether it's CAD or other things that are coming along, the ability to transfer information around and data to information and the information is the patient care. So I think the way we do things is really going to change, really have to change because if not you might as well get 50 slices.

QUESTION: Congratulations for outstanding work and excellent presentation. You mentioned about the issues of the high resolution. So using this technique looking at dynamics of the contrast along with all the vascularities, how confident are you in making diagnosis before biopsy or looking at tissues later?

FISHMAN: I'm always confident. I think what happens, they always like to say is that pathology is something additional that spends money for no reason.

[laughter]

FISHMAN: But I think, you know if you look at i think radiology now better than ever is able to look and be very, very, specific. I mean 1 of the things we learn from the pancreatic conference and multidisciplinary conference is pathology is very definitive. It ends up the reading of slides varies by 30%, so in radiology this does not look ache pancreatic cancer, it looks like lymphoma or something else, you go back and do staining so maybe more and more is going to signatures of lesions. I think the old days when we scan, you know 5 or 10 millimeters and there was no ability to scan quickly. You didn't have that information. Now we do, and as scanners get faster, I think we're really able to create these signatures and these maps. I think that's going to be a very powerful tool.

QUESTION: yes our question regarding the lesion of the vascularity and tumor in the vasculatures for a thrombus? In terms of, you take the ooh bc and it has the filling effect, thrombosis verses tumor, so in renal cells for example, if the thrombus is vascular then it's a tumor. Sometimes you can't be sure, the filling may be a thrombosis but if it enhance, it would be easy but if doesn't then you can't say for sure.

BLUEMKE: I want to conclude at that point. Thank you very much for presenting the Doppman lecture. I would like to present this on behalf of the NIH in behalf of the Doppman memorial lecture series. Thank you, Elliott.

[applause]

BLUEMKE: Thank you and we'll conclude there. There's a reception in honor of Dr. Doppman and Dr. Fishman in the conference room that everybody's invited to. Thank you.