CT Technology: The Early Years - Norbert J. Pelc, ScD
so I'm very pleased uh to introduce to you our next speaker Dr Norbert poulk began his research in CT in the mid-1970s and after graduate school he worked on CT and other modalities at General Electric Medical Systems and later joined the Radiology faculty at Stanford University he's a member of the National Academy of engineering and the National Academy of inventors and was granted an honorary doctor of Medicine by the Friedrich Alexander University of Airline and Nuremberg he's a fellow of the aapm the ismrm the spie the AHA and the aimbe his Awards include the outstanding researcher award from the arsenea and the Edith H Quimby award from the aapm and I look forward to hearing his presentation over thank you Cynthia for the invitation to come here and speak today and the opportunity to see so many good friends um as we've heard the first CT scan was done in the UK in 1971 October of 71 and the first CT scan in the United States was done here uh 50 years ago this talk uh will cover some stories I'm not going to go into technology in detail rather try to tell some of the stories behind the developments um there were there was a lot of early research related to CT we often nowadays start the description of the technical aspects of CT with radon and work that was done in mathematics nearly part of the 20th century there was also worked on an electron microscopy and and radio astronomy uh later on we learned that there was research done in Kiev in the 1950s on on x-ray CT but the the people who contributed to the story that developed the scanner that we're commemorating today really didn't know any of those stories or that technology so we will start our story today with Alan cormack uh I want to acknowledge the many sources of of information and a few things that you could look at if you want to read a little bit more about this um I want to give much of the material comes from two papers that you see listed on the bottom uh and I want to give special thanks to the co-authors that contributed to those two papers Raymond Schultz who's who's here in the audience uh Jay Stein and and David chessler and then also listed you see a number of people who have shared uh their memories and and materials with me uh for this talk contributed to those papers as well so we will start our story with Alan cormack as I said uh and then uh we'll talk about a very early work that was done at Siemens before the development of the first uh of the Emi scanner before introduction of the Emi scanner I should say um and and as we do that I want to talk about the mindset of uh people developing technology and using technology and radiological Imaging back in that day because that affects some of the things that happened we will uh then talk about uh early CT related work at MIT and Mass General Hospital before the the first use of the Emi scanner and then talk about hounsfield and the development of of the scanner that was eventually deployed here we'll then talk about uh other Market entrants and what happened in the radiological field especially on the commercial side but not entirely uh commercial uh in the period of Rapid innovation on petition cross containment you heard a little bit about that and Market consolidation we'll talk uh and then some closing comments uh Alan cormack was a South African physicist who worked part-time in radiation therapy and uh when he thought about how would one go about doing accurate treatment planning he realized that what we needed was three-dimensional maps of linear attenuation coefficient uh and so the thought came to his mind of how would one produce such images from measurements of transmission measure you know made externally and that you needed to reconstruct a function from those measurements and he did some literature search and did not find radon uh and didn't didn't find anything on how one would reconstruct this was in the in the mid-19s early 1960s um so he did that literature search was so surprised to not find a solution so he came up with his own and then uh approved that it worked by making measurements of a phantom with a Cobalt 60 source and reconstructing the attenuation coefficient distribution of that Phantom he then approached the physicists at his hospital and told them about this and uh to no avail nobody was very interested in this the story that I want to tell you about Siemens is is told in that paper that you see listed on the bottom by by Walt Robb um an R D team at Siemens in the uh 1960s realized uh led by Professor guten realized that they Could reconstruct a map of linear attenuation coefficients from external measurements uh the Radiology field at the time was really focused on improving spatial resolution feeling was the way that radiology would be better is if the images could be sharper that extra detail would allow them to see the UN unseen things and so that was the mindset and so these scientists at Siemens felt that uh the technique could only be useful if they could match the spatial resolution of radiography uh so say on the order of 150 Micron resolution and when they looked at computer technology that was available and how it might progress They concluded that sufficient computer technology wouldn't be available for decades so the project was stopped nothing was published as far as I know or patented at least I couldn't find any by this time hounsfield was already uh well underway uh with developing the Prototype scanner I want to talk about a couple of other people who were working uh in the CT area that uh before the the the first demonstration of the of the Emi scanner one is OLED tradiac he trained as an electrical engineer at MIT and then joined the faculty there and then the 1960s worked at MIT on image processing and and image recognition with interest in biomedical Imaging he independently derived reconstruction by Fourier interpolation he actually independently as far as I can tell independently derived what we now call the projection slice theorem or the central slice theorem and he made cross-sectional images of an excised bone from scanned radiographs of that bone on a turntable as the turntable was moved he went to one of his colleagues in the ee department at MIT and told them about this and as it turned out that person had just read a paper on electron microscopy Reconstruction from projections to the paper would have been that de rochier include paper and so trediak felt scooped because someone else had just demonstrated this reconstruction by Fourier interpolation it makes you wonder what might have happened if he realized that the images that he had produced were one of the first uh CT reconstructions of linear attenuation coefficients made by projection measurement he didn't know that Carmack had already done that as well David Chesler uh is a mentor of mine and also Steve readerers Dave chessler also came out of the MIT ee program and worked in industry and joined Mass General Hospital in the late 1960s because he didn't want to work in the defense industry he instead worked in nuclear medicine and independently derived filtered back projection reconstruction for with with his interest being for the the use of that reconstruction and emission tomography with single photons and positron imaging which had been developed in that laboratory at Mass General Hospital however in the process of producing these images he wanted to do attenuation correction so for attenuation correction of those images he also acquired transmission measurements uh and reconstructed images of linear attenuation coefficients those would have been very very noisy um uh but nonetheless they probably were some of the very earliest reconstructions of linear attenuation coefficients uh that said clearly after Cormac oncefield and trediac as hounsfield had by then produced images in his laboratory those results were presented uh in 1972. so now let's uh talk about the story of hounsfield some of this you saw in the movie uh he um trained uh and got an associate's certificate that's the extent of his formal training he during World War II worked on radar and then joined Emi and worked on a variety of topics including radar system displays general purpose computers and pattern recognition and as you saw in the movie also in the 1960s he was pondering uh various things and he asked himself could he uh recognize text in a closed book by making external measurements or could he discern what's inside the Box uh from measurements made uh readings through the box and you you saw that told in in the movie on a holiday as they would say rather than on vacation on this holiday he met a physician who bemoaned the drawbacks of radiography and the difficulties in in seeing things in radiography uh because of superposition and so hounsfield realized that that is an example of his what's unknown thing in the Box how do you reconstruct what's inside the box from measurements made from the outside so he started working on that a little bit more and he considered uh the problem of nine unknown numbers or unknown numbers in a three by three box could he come up with a method of reconstructing the contents of those nine cells from measurements made at 0 45 90 and 135 degrees and uh and he came up with what we would Now call the algebraic reconstruction technique um he then worked with a colleague he wasn't apparently much of a programmer and neither was carmacked by the way uh and he worked with a colleague who who programmed that and was able to produce images of an eight by eight Matrix uh from a version of that in 1968 he approached Emi for funding for the idea of making a medical scanner uh using this principle and Emi decided that they would not allocate internal funds until he got substantial funding from the UK government and so he applied for uh funding from the Department of Health and Social Services in the UK and we're all familiar with this his Grant application was rejected so he resubmitted that Grant application in October of 78 of 68 I'm sorry and uh you know it's great that some people preserve these things because we still have copies of that Grant application and you saw some images from that Grant application in the movie as well um it was called an improved form of x-radiography and in the introduction of that Grant proposal you see the text that I outlined here if the object to be studied was one such as a book normal methods of X-ray pictures so you see the connection between the Grant application and his uh his the the problem that he was trying to solve of uh Discerning what's inside of a box from measurements made from the outside so in that and this uh the image on on your left uh you also saw in the movie which is one of the figures from that Grant application that shows the idea of an x-ray source and detector that would scan across the patient and then rotate and make measurements one slice at a time but on the right you see another figure from that 1968 Grant application that shows cone beam CT with a 2d detector um so in that same Grant application he described what he foresaw as the potential benefits of this technology high dose efficiency sensitivity to minute changes in in attenuation lack of superposition uh and uh envisioned a possible role in the detection of cancer he also foresaw that we would need some way of displaying these images which would have a very wide dynamic range much wider than any medical image at the time uh and uh that this is how window level uh idea uh came into radio radiology it was not available before that because there had been no need for it or at least no recognized need uh in the Grant application he discussed a variety of technical topics like sources detectors scatter and beam hardening and the relationship between scam time and various system design choices to achieve his goal of this uh one percent uh of the attenuation coefficient of water resolution and density so the grant was uh approved and the combined funding is combined funding from Emi and the UK government was 69 000 pounds and that led uh within a fairly short time to the first CT scan of a patient uh in about three years and as you saw about uh a little more than a year later the first CT scan here in the United States at Mayo Clinic the first uh scanner that he produced was called the lathe beds scanner and it had a radioisotope source rather than an x-ray source so the scan time with this machine was nine days to get sufficient uh photons so that the americium source was replaced with an x-ray tube and the scan time came down to nine hours and uh they required at the time two and a half hours of CPU time on a Mainframe to produce a 32 by 32 image they scanned brain specimens and a variety of other things like the pig carcass that you see there Ambrose who was a neuroradiologist who was working with Ambrose and providing with uh hounsfield and providing advice even with these very crude images recognize the potential that they would have and and said to hounsfield you realize what you have done so in uh January of 1970 Emi decided to produce uh and build a prototype uh while this was happening uh hounsfield visited a number of Radiologists and uh and uh told them what they they were trying to do at Emi and in general got a very cool response uh and I think in large part it was because of the the spatial resolution uh which which was going to be on the order of three millimeters uh far far poorer than radiography plus this machine would require computers which were not being used in uh to any extent in Radiology the machine would be expensive and in general I would say that they didn't realize what the impact would be of the far Improvement in density resolution that would be gained by this and the removal of the superimposition nonetheless by August of 1970 the specifications for the scanner were complete uh the machine uh you see the photograph you see there is the the Emi mark one which was that first prototype that was installed at Atkinson Morley hospital and Wimbledon London it had a water bag that patients had was in a water bag and the water bag was filled with water and that had a number of technical benefits and as you saw in the movie It produced two slices in a single scan of about five minutes used a fixed anode x-ray source and the detector was two sodium iodide crystals each with its own photomultiplier so as a two slice scanner even though we that then didn't call them multi-slice or multi-detector row didn't have multiple rows just multiple I don't know can you have a row of one maybe the data raw data Matrix was 160 samples and one 160 views uh they were written on magnetic tape the Magnetic Tape was then taken back to the Emi Central lab where they were reconstructed on the Emi Mainframe an 80 by 80 images were printed on a Polaroid film and and returned about two days later uh as I mentioned the first scan was performed in October 1st 1971 uh and and the image you you've seen before it's often uh displayed of a patient with a cystic astrocytoma um Ambrose recalls that he and hounsfield jumped up and down like football players who had just scored the winning goal uh with the the success of that initial scan there was a initial presentation of the results from the Emi scanner at the British Institute of radiology conference uh in April of 72. and also as you heard in the movie there was then a presentation at a neuroradiology postgraduate Course in New York City in May of 1972 that generates generated huge excitement because the leaders in neuroradiology in the United States were there and they uh quickly appreciated the potential that these images presented uh among the things that were retained um and actually when I was speaking with Steve last night I recalled reading reading this particular quote that I inserted in the talk uh last night specifically for this audience but there were notes that were taken and and memos that were written uh by the Emi speak people of uh the reaction at this uh Course in New York and also some visits to potential customers in the United States that hounsfield and some of the in my people uh performed immediately after that that uh course uh so here's uh one Emi manager noted that hospitals like Mayo Clinic expect to examine 30 patients daily the Mayo Clinic will examine as many patients in one week as had been examined by accidents in Morley Hospital Wimbledon during a whole year it's almost correct right it's certainly within a factor of two oh uh Mayo Clinic was mentioned in those early internal memos by Emi people uh so shortly after that the Emi scanner was shown at the rsna in 1972 uh there was there was no planned uh scientific presentation uh at the time because all of this happened too late for the submission deadlines and I think it's fair to say that in my uh as a corporate entity didn't appreciate the importance of the rsna uh and uh what what could happen as a result but they learned of that and they had an exhibit in 1972 and Ambrose and hounsfield were able to speak immediately after the president's address at the rsna meeting that that year and it proved to be a huge success and major U.S hospitals started to order machines and what we learned is the Mayo Clinic essentially bypassed that there's a a Serendipity that I think is worth noting the emi's initial estimate of the market for for CT scanners was 12 machines worldwide and part of this was that the model of reconstructing images on a Mainframe and returning them was just really not viable and what enabled CT to become a reality and to be distributed so quickly to so many places was the development of the mini computer so just before this was going on in the mid-1960s many computers started being introduced and by the time they were needed for CT scanning we had availability of sufficiently powerful mini computers at reasonable price and they got incorporated into the scanners and the reconstructions Could Happen locally and the data wouldn't have to be sent and so here you see a photograph of uh emi's scanner I believe this was at the rsna exhibit in 1972 and uh over here on the right you see the data General Nova mini computer that was the brains computational system behind it so what was the industry reaction I mentioned this uh paper by Walt Robb on his Recollections of the first 10 years of the CT industry and uh so he was there at the time uh he was the uh the head of GE Medical Systems as this was happening uh and he said most manufacturers of radiographic equipment concluded that the this new scanner would be a niche product that it wouldn't uh greatly affect the field uh and this was in part because of the long scam time four and a half minutes uh and uh and the modest spatial resolution uh that was achievable at the time Siemens uh as you recall Siemens had uh looked at this idea and in a sense set it aside um uh Professor guten who was uh leading that who had led that team at Siemens was astounded by the fact that uh these images which were proving to be so powerful only had an 80 by 80 Matrix about a 30th uh in each Dimension compared to what they thought would be needed and then uh this is also from Walt Rob's paper one member of the team confessed it took us months to comprehend the significance of density resolution versus spatial resolution and now we had missed been misled by being too closely associated with the conventional x-ray business and and I think that is really uh very perceptive uh and you note that these the idea of of CT and uh achieving uh direct measurement of linear attenuation coefficients in this 3D way came from outside the medical imaging field by people who were who were not tainted by this bias of so at Siemens the internal project as you may not be surprised was re-initiated uh and and Siemens embarked on the development of a CT scanner I want to digress a little bit and talk about a story that that pertains to my own experience and also Steve reader I I mentioned David Chesler who was at Mass General and had developed reconstruction algorithms for nuclear medicine the head of the laboratory was Gordon Brownell who was a Pioneer in what eventually got called positron emission tomography and that was really the main activity of the physics research lab at MGH at the time but in mid-1972 I want you to think about when that is mid-1972 there is no CT scanner in the United States we're still a year away from installation of the scanner here uh they're the only scanner that had been made was the one that was uh in London and the only public presentations were at most a handful including the one in New York but Gordon Brownell decides that the laboratory which was really fully involved in positron imaging would develop a whole body x-ray CT scanner uh and uh they uh started on that project Dave chessler was the lead person and uh eventually in a Grant application was submitted to the National Cancer Institute to develop a whole body scanner uh the submission was in 1973 and the grant was funded in November of 1973. uh roughly at the time that the uh first scanner was installed at MGH but by then the concept for this body scanner had already been developed and this is an artist's rendering of what the scanner was going to look like it was a third generation scanner rotating fan beam 10 second scan time so that it could do body Imaging the detector was a 256 element xenon ionization chamber aligned with quarter offset re-binning to parallel rays these are all things that we're familiar with nowadays one unique thing about this scanner is the channels in the detector array were unequally spaced in so that when we've been to parallel rays they were equally spaced and you didn't need uh interpolation in the radial Direction in the reconstruction so both Steve and I were involved with this research another activity by the NIH was in 1974 the NIH issued an RFP a request for proposals for development of whole body CT scanners there was a competition for this it was won by uh Columbia Presbyterian Medical Center uh and the Neurological Institute of New York where the pi was sadak hello and uh Raymond Schultz was involved uh in that program and they collaborated with two subcontractors including American Science and Engineering which was based in Cambridge Massachusetts uh in general I would say the federal investment in this early CT work was impactful not just for the research that was done directly by it but for the atmosphere that it set because the companies got involved and started investing in CT in part because they wanted a piece of that uh action right and a number of companies submitted applications to this RFP which as I said was won by this team with sadek Hillel and American Science and Engineering we'll come back to a piece of that story in a little bit the commercial activity then begins in 1974 right after the introduction of the Emi scanner the year before Siemens and Hitachi introduced first generation head-only CT scanners uh the same year Robert ledley who was uh at Georgetown uh founded a company he called disco and they developed a uh body scanner which was essentially a scaled up version of the Emi headscan uh it uh five minute scam time first generation translate rotate very few uh of these were sold in 1975 Emi introduced the seat the 1010 which was a second generation uh and those of you in the field known in second generation we have multiple detectors scanning and rotating making the scanner faster so it it was eight eight detectors uh and the scan time was reduced to one minute and it also had no water bag the same year Ohio nuclear introduced the Delta 50 head-only scanner in the Delta I'm sorry Delta 25 and the Delta 50 body scanner neuroscan which very few have heard of developed a head-only second generation scanner I know that because GE uh was marketing that scanner I don't think they sold any syntax uh maybe this was the first drug company that got into Medical Imaging they were followed by a number of other drug companies later that got into the medical imaging Market introduced a head-only scanner and the same year the Emi 5005 that I think you heard talked about uh by uh Dr Sheedy uh was introduced and it had a second generation Scanner with the 10 second 10 degree detector Arc it could scan in 20 seconds if you think about the idea of this mechanical Gantry scanning and rotating and scanning and rotating and being done within 20 seconds that's pretty much as fast as you could get uh translate rotate scanner to go The Quest For Speed was on though because people realized we needed faster machines to scan outside the of the skull the concept of third generation scanning is often credited the Doug Boyd but it seems to have been uh in the air as I mentioned Dave chessler had come up with the same idea as had hounsfield um uh the scanners were built for Speed but the first two third generation scanners that were built were not for body Imaging the first one is you see here was the gectm it was a breast scanner and the first scanner was installed here at the Mayo Clinic the uh I mentioned relevant to what uh Dr Sheedy talked about of people putting things in the head scanner it was motivated by somebody at Mayo I don't recall the name but there are Publications to this effect putting mastectomy specimens in the head scanner to get a feeling for what breast tissue would look like and you see there in the photograph red Reddington and Lonnie edelheid some of you might know the same year artronix introduced ahead only third generation scanner later that year GE introduced the body Scanner with faster scan time artronix introduced a body scanner Varian introduced a body scanner you see the photograph of it there and it was unique in a number of ways including the fact that it had a slip ring and and had continuous rotation at only three seconds uh and all of these were third generation scanners with Xenon ionization Chambers they were this detector technology is one that physicists trusted for stability and linearity Siemens was bucking that Trend Siemens produced the first scintillator photodiode detector in a machine that was called the somatome and in that machine they they uh had had another interesting idea you see here in the bottom left insert it had two detector arrays one for the head and one foot for the body that shared the same Electronics so you could uh put your 256 elements in different size field of views Phillips did this in a different way by having variable magnification in a third generation scanner that could change the location of the source and detector all the third generation scanners at the time produce ring artifacts you see examples of that there because of the uh stringent requirements on detector performance J Stein who was at American Science and Engineering recognized this problem and new detectors well enough that he was very much afraid of trying to beat that requirement and he came up with the idea of fourth generation scanners in order to avoid ring artifacts and this was part of that program that was headed by sadaqilo It produced the asnd scanner in 1976 that had a five second scan time and in my recollection It produced the very best body Images at the time uh it also had post patient collimation on each of its 600 detectors so as to be able to achieve higher spatial resolution at the cost of dose efficiency the fourth generation design was later adopted by two other vendors technicare and picker Emi uh came up with a hybrid version of fourth generation scanner they wanted the source to be on the outside of the ring and so the Ring of detectors had to mutate and get out of the way when the X-ray beam was penetrating through that but the most uh a strange machine was a fourth generation scanner by uh artronix and they liked gas detectors so now try to come up with an idea of a gas detector that has internal collimation and the cells are very directional how do you put that together with a fourth generation design The Source has to be in the center of a ring of detectors which you see here but then the patient can't be in the center so here you see the patient off to one side and then the entire thing rotates about the patient as the source rotates within the center of the circle I remember seeing this machine at the rsna and just was astounded and there was a movie of it so here's the Ring of xenon ionization Chambers the source is here the source is going to rotate around this as the uh Circle wobbles about the patient opening very ingenious I thought and it's a slip ring scanner it could keep doing this all day long I don't think more than a handful of these were sold okay I can't Advance it good uh some other technology that was developed at around the same time the Electron Beam CT scanner came out of UCSF an academic center and then was later commercialized by imitron targeted at cardiac imaging and recognizing the need for Imaging multiple slices with very short scan time and of course we have the work that was done here on the dynamic spatial reconstructor you see a photograph there and a and a drawing of it it had multiple luxury sources it was a inherently a 3D volumetric scanner um and could produce snapshot images of the heart you see examples of that here from a paper that came out of the group here at the Mayo Clinic um uh really quite Visionary in many ways because uh inherent in this view of what CT of the heart should be was an understanding that it had to be three-dimensional it had to have very high spatial resolution and you needed to have the ability to reformat it to the different planes at which you would want to view the cardiovascular system so getting back to the competition GE had received advice to drop third generation design because of ring artifacts and and the pressure that was being put on the market by the ASE scanner but they resisted that were able to solve the ring artifact problem and then developed a higher resolution scanner that was called the 8800 um other third generation companies soon followed and that this puts tremendous pressure on the fourth generation scanners because they had an inherent cost disadvantage and other dose inefficiency and Scatter problems show third generation scanners began to dominate the market uh at about the same time and you heard from Dr Sheedy um and and Dr Forbes the the government imposed uh cost constraints uh and this uh had an impact uh as did the introduction of MRI and here you see uh CT scales over time and you see a peak and then the introduction of certificate of need and the size of the CT Market drops it then recovers and then it's hit again by uh further government regulations as well as the introduction of of Mr this was especially hard this reduction in Market size on the small companies and the small companies began to drop out so early on this Innovation came from not only outside the medical imaging field but certainly outside of the major Radiology companies but uh by the early 1980s all of those Market entrants had disappeared and the CT Market was dominated by the traditional Medical Imaging companies in uh 79 the Nobel Prize is uh given to Cormac and hounsfield for the development of CT um and in the award it said it's no exaggeration to state that no other method within x-ray Diagnostics within such a short period of time has led such remarkable advances in research in a multitude of applications some closing thoughts uh before CT as you heard uh x-ray Imaging was in was focused on improving spatial resolution there was a bias against accepting any any lower spatial resolution even if it obtained true 3D image it's a very good density resolution this Innovation came from Outsiders only now we're approaching the spatial resolution that Siemens thought would be necessary uh we're getting getting there government funding for the uh This research and development was really instrumental uh but then government intervention in the market was disruptive and and perhaps negatively so the introduction of CT brought computers into uh radiological Imaging I don't know if we're happy about that or not uh and the success of CT paved the way for other Innovations including magnetic resonance imaging thank you
2023-07-22 17:51
