|
|||||||||||||||
|
Home
|
A Radiology Today
Interview with Robert L. Falk, MD — 3D Image Processing Radiology Today (RT): You’re a practicing radiologist and cofounder of the first purely commercial 3D image processing laboratory. (A few institutions have developed in-house labs and subsequently offered their services to the outside, and teleradiology firms are beginning to offer the service.) Just what does 3DR Inc. offer as an advanced image processing laboratory? Robert L. Falk, MD: We offer a combination of hardware, software, and services, with the emphasis on the service end of it. We don’t really sell anybody a workstation. We bundle the cost of the workstation into the cost of a package of hardware, software, and volumetric reconstruction cases. The core of our business is doing the reconstructions for the customer. We also provide the infrastructure for that to happen. We don’t sell hardware and software and then go back and do annual licensing fees and service programs; all of that is provided. We basically work more like a cell phone provider. You bundle in the cost of the phone with the cost of how many minutes per month you use the phone. We’re doing the same thing only it’s reconstruction cases per month and a server configuration that makes sense for our customers, whether it’s four concurrent users on a server, eight, or whatever. RT: At this point, how many hospitals are doing volumetric reconstruction? RLF: I would say that most hospitals are doing some level of advanced image processing, even if it’s just some multiplanar reformats and basic surface rendered views produced by their scanner. Many scanners now include some fairly sophisticated 3D software. From what I’m seeing here in the Ohio River Valley, virtually every hospital is doing some degree of 3D work, but they’re very early on in the process. One year ago, I would have told you that the majority of hospitals were not doing 3D work. In the past year, I think we’ve tipped the scales a little bit. Most places are dabbling in it and doing some of it. And some places are doing a lot and are obviously quite sophisticated. Hospitals and referring physicians are starting to see the need for it. When they replace their 16-slice scanners with 64-slice scanners, they often see the need to use 3D for common CT applications. Coronary CTA [CT angiography] is a different animal. RT: What’s different about coronary CTA? RLF: A minority of facilities are doing coronaries. There are a lot of political issues involved that scare administrators and radiologists about cardiac. There are going to be turf issues and political issues in cardiac that are going to be stronger than they are in the other areas. RT: When did 3DR get started? RLF: We officially started signing on customers in the fall of 2005. We purposefully kept ourselves small. For about a year, we worked with four or five customers as we worked this thing out. We felt like we were in the market early. There weren’t a lot of 64-slice scanners out there at that point. Last year at RSNA, thin client started to hit. That’s another aspect of 3D growth. We decided to switch over to a blended laboratory with a thin client setup rather than just having a fat client and actually bringing their images to our laboratory. We partnered with Mercury Computer Systems to be an OEM [original equipment manager] for their Visage CS product and get a full functioning thin client server out in the marketplace. What we now have is a blended laboratory offering a combination of laboratory services, workflow management solutions, and technologist education. We can put a server in a customer’s facility, have our techs reach out to that server through a VPN [virtual private network], and actually function as if they were the facility’s own techs in their own department, even though we’re in Louisville [Ky.]. If it makes more economic sense because the customer is smaller, we can actually lease time on our server in our laboratory to the customer at a lower rate. What happens then is the images are transferred to us, we do the 3D reconstructions and the images are pushed back to the customer’s PACS from our server. The customer has the ability to read the case as reconstructed by 3DR or to go through the VPN and manipulate them further on our server. And we also employee the tradition dedicated workstation system. We use Vital Images’ Vitrea workstation in addition to the Visage product to manipulate the images, and the reconstruction gets integrated back into their PACS ready to read. With that system, the customer doesn’t have the ability to do any further manipulation. RT: The thin client server approach is where this is going. The idea that you, the radiologist, will order the reconstruction that he or she wants to see and that’s going to be the end of it is quickly going away. RLF: That is going away in the cardiac world, but some things that lend themselves to specific protocols and other exams don’t. It really depends on the study. I think, for example, that with renal CT angiography, a technologist can put together a group of five or six series that is going to give you everything you can do with renal CT angiograms. There is not a lot of variation 99.9% of the time. And you can do it in the orthopedics world. You can usually get the money shots that summarize in as concise a fashion as possible all the relevant clinical information to help the radiologist or, just as important, the ordering physician. RT: So most radiologists will want the techs to produce most images but still give the doctor the flexibility to jump in and work with the image as needed? RLF: In my practice—and other radiologists do things a little differently—radiologists do not do a lot of postprocessing. Until about one year ago, I think I was the only guy who knew how to turn on the workstation. We rely on technologists to do the bulk of the postprocessing. We like to sit down at the PACS workstation with a worklist. What we have built in our practice is a strong emphasis on the universal worklist. We want to move images and not radiologists because we work from multiple sites and we’re subspecialists. If we have someone who should be reading the MRIs, we want that person reading all the MRIs. We really stressed centralized PACS with redundancy, and we’ve built our workflow around that. We’ve built our own in-house laboratory. My hospital is not a customer of 3DR’s yet. I don’t have any conflict that way. We are not as completely centralized yet as we would like to be. We have workstations in two or three different places, but it is all integrated into the same PACS. If I’m reading CTs, I may read a chest CT PE [pulmonary embolism] protocol, a CT of the brain, an abdomen/pelvis, a coronary CT angiogram, and then a carotid angiogram. It’s just one case after another. Nineteen times out of 20 what the tech has created for me gives me enough information to read that case. In the odd case where I need more information, I’ll call and say, ‘On Mrs. Jones, I’d like such and such a view, and I’d like to see such and such a vessel.’ The tech will work the image a little differently and put it back on the PACS. If the processing requires something that I don’t think the tech can handle or would take me 10 minutes to explain and 5 minutes to do it myself, I’ll sit down at the workstation to make the picture that I need to answer my question and move on. That’s our workflow. Our goal in my group is to create a 3D case as similar as possible to a routine 2D axial case. RT: Does that change with a cardiac CTA case? RLF: It changes some in that the radiologist has to jump in a little more often. Right now, we don’t do that huge volume, and we handle it the same way. We still have the techs create a set of reconstructed images, but it often requires the radiologists to go over to the workstation, much more than for a renal or carotid study. RT: Radiologists doing
more at the workstation seems to have become a driving force for the
thin client idea. In cardiac cases, you’ll frequently want the
access to manipulate images yourself and a system that delivers that
capability. In my opinion, the ideal environment is to have an extremely well-trained tech who will present to me images of outstanding quality that will allow me to do 90% or more of what I need to do to interpret images. If I need to jump in and answer a question that wasn’t answered, then I need to do that quickly and get back to doing what I was doing before. RT: For the noncardiac 3D work, the tech provides the pertinent reconstructions and axial slices and you just read from the whole set? RLF: Most of the time, that is more than we need, but we design it that way. RT: More studies with more images means reading from volumetric reconstructions is only going to increase. Won’t radiologists learning to work in this environment probably like seeing what they’re used to along with what they’re learning to use? RLF: It is an incremental move to a dramatically new technology. 3D image processing is a way of looking at pathology. Back when I was a fellow doing MRI when we had just gotten our first MRI system, the attending physicians would look at the MRI and then say, ‘Well, let’s look at the CT,’ because that’s what everyone was comfortable with and the MRI was so new. There was a level of comfort to see the CT and then move to the MRI. It’s the same thing here. There is just a level of comfort seeing those axial slices, but as time goes on, you end up relying on the axials less and less and the reconstructions more and more. There is a time period where you train your eye to do that. RT: From your customers’ perspective, how does 3DR work in an operational and financial sense? RLF: We try to make advanced image processing revenue neutral for our customers. The 3D end of it will never be a moneymaker for a hospital. The reimbursement is too low and the time factor and infrastructure, both in people and equipment, is too high. Building your own 3D laboratory in-house is wonderful and a great thing to do—just don’t plan on it becoming a profit center for your department because of what it takes to maintain it. Purchasing 3D services from us is very much like buying cell phone minutes. If someone thinks they’ll need seven or eight concurrent users at some point, we’ll bundle an eight-user arrangement with the total number of reconstructions they’d like us to do for them. Obviously, the more volume we do for them, the less we charge per case. RT: So the radiology administrator projects that her facility will need so many reconstructions per month, and that they’ll cost so much per exam, and how they want to transport and process the images. Then she projects reimbursement rates and does the math and, when the number seems right, signs a contract? RLF: Basically that’s the way it works. If the customer is buying 100 cases per month and is only using 20, it’s our responsibility to go back to them and rework the arrangement in a way that makes more sense because clients are buying more capacity than they need. RT: So you don’t do rollover minutes like AT&T? RLF: (laughing) No. We don’t have rollover minutes or a family plan yet, but we’re working on it. Basically, we can be somebody’s 3D laboratory for a few cases per day. Let’s say a hospital does its own reading during the day but wants us as a backstop for busy times, nights, and weekends. We can put in a server that would make every PC in the facility a fully functional 3D workstation and distribute that technology throughout its entire enterprise. We can do that and give it 24/7 coverage for roughly the cost of one CT tech per year. If you do it yourself, you have to hire at least one tech—and probably two—and pay them to be on call, as well as all the costs associated with that. We can handle all your 3D if you want us to or just night and weekend call. And since it’s a subscription type of deal, the cost comes out of your operating budget, not your capital budget, allowing you to budget month to month and know what your 3D lab costs will be. 3DR helps facilities do 3D without putting the capital—not only financial capital, human capital—into doing it themselves or building it as a 24/7 proposition. RT: How is volumetric reconstruction reimbursed? RLF: There is reimbursement for advanced 3D postprocessing. There are basically two levels of reimbursement through Medicare CPTs. There’s the 76377 code, which pays roughly $120 to $200, depending on the location. That is for advanced 3D processing performed at a separate workstation under physician supervision. An example would be 3D processing for surgical planning for resection of a bone tumor. That needs to be done in 3D. Then there is the 76376 code, which pays substantially less than that, roughly $40 to $60, depending on the region, and is meant to pay for renderings that the modality produces that do not need separate postprocessing. An example would be the PE protocol chest CTs that are being done now with coronal and oblique coronal reformats. They are done at the modality console by the CT tech. RT: You mention the flexibility of outsourcing all or part of your 3D work. What should imaging facilities be considering when making that decision? RLF: You can reasonably do some advanced image processing in your hospital Monday through Friday during the day. You can get to that by training one or two techs. I think it is unreasonable to expect all your CT techs to do postprocessing work in 3D, and I think at some point it becomes a workflow and financial issue. When you have a 64-slice CT scanner that scans a patient in 5 seconds, the rate-limiting step in how many patients you can scan per day ceases to become the speed of the scanner. It becomes how fast can you get an IV started, how fast you get the patient consented and comfortable on the table, and all the information typed into the PACS. The limiting factor becomes all those ancillary things that lead up to the scan. Those things are people-intensive, and the number of people you can devote to the CT scanner is directly proportional to the number of people you can scan in an hour—and that’s directly proportional to your bottom line. If you pull a tech from the scanner environment and put him or her in the workstation environment, your [scanner] productivity takes a hit. Can you afford that on a good day? A bad day? On a day when your best tech calls in sick at 2 o’clock in the morning? What we hope to be at 3DR is a workflow buffer. We may do 100% of the processing in some places and 10% in others. There are people who may only use us on nights and weekends and some that may use us at certain times of the day when the crunch hits and they must devote all their techs to scanning patients. We give them the ability to manage the workflow literally hour by hour as opposed to having to staff shift by shift. … Every hospital has different needs and is going to use us differently, but they have the ability to make that decision. RT: Radiologists transitioning to 3D interpretation is obviously one reason you use it, but if my facility isn’t doing cardiac CTA and the doctors don’t have a particular interest, why get involved? RLF: I’ve lived through this with this company; with each client we actually have four types of customers. There’s the radiology department administrator who we actually have contracted with from the standpoint of service and turnaround time, HIPAA, and all those things. We have the radiologist who has to be able to interpret the study we give them. We have the referring physician who has to be able to sit down and use that information to help take care of the patient. And our fourth customer is the patient. We’ve found that radiologists can pretty much take some multiplanar reformats, a few curved reformats, and the raw data images and interpret the study. As radiologists, we’ve spent our whole lives taking 2D images and, in our brain, converting them to 3D. Surgeons don’t really work that way. Primary care docs don’t work that way, and patients certainly don’t work that way. I have a good friend who is an internist here in Louisville. I once asked him how he goes about explaining things to his patients, such as a patient who has a renal artery stenosis that he is going to send him to an interventional radiologist to have a stent placed. He told me, ‘I probably spend 25% to 30% of my day drawing pictures for my patients in the exam room. It’s amazing how much time I spend drawing pictures.’ One advantage of multislice CT and MRI with finally having good 3D software available is the ability to create these images. Understanding the disease process is intuitive, and I think patient education is going to be a fairly substantial beneficiary of advanced image processing. Doctors are going to like that, too. RT: If it is valuable to the referring doctor, it’s going to be valuable to the radiologist and facilities. You hear at meetings that one challenge facing radiology these days is the sense that if other doctors are just reading images, how do radiologists remain relevant? One way may be by providing images a referring doctor can share with his patients and say, ‘There, see the block? That’s what the interventional radiologist will go in and fix with a stent.’ That would be an example of the added value beyond just making the diagnosis. RLF: That’s right. In radiology, you have a lot of different people who you have to satisfy. As far as building physician loyalty, it is a competitive environment out there. The ability for an imaging center or a smaller or midsize hospital to offer services that larger facilities offer without building the entire system is appealing. The ability of patients to get specialized studies is appealing. It really is a good way to level the playing field, giving hospitals quality 3D services from day one. 3D Rendering CPT Codes 76377: 3D rendering with interpretation and reporting of CT, MRI, ultrasound, or other tomographic modality; requiring image postprocessing on an independent workstation — Sources: American Medical Association and GE Healthcare
|
|
3801 Schuylkill Rd • Spring City, PA 19475 Publishers of Radiology Today All rights reserved. |