SIIM in 3D!

Due to various reasons, a new puppy being first and foremost, I was only able to make it to this year's SIIM for one day. Fortunately, I was able to maintain my dignity, amidst a week's worth of house training, and I caught myself before praising any product with the phrase, "Good poo poo!"

Sadly, I didn't get to see as many toys as I had hoped, and I didn't get to stop to see several of my friends in the business. I'll have to rely on others to tell us, for example, if PACS-IW is available yet (rumour on the floor was that the integration to IDX, I mean Centricity RIS-IC still doesn't work, but that is as yet unsubstantiated).

The images above show a the race-car-shaped USB key (and its instruction sheet) given to all attendees, containing all the documentation of the meeting. No tree-killing paper syllabus for us! Of note is the wonderful admonition to:

Be careful when unplug USB Flash Disk. If USB Flash Disk is unplugged suddenly while data is being read or wrote, data may be lose.

I guess I don't always speak the King's English either.

I have to wonder what the carbon-footprint of manufacturing hundreds of little USB cars might be. . .note the logo on the roof which tells us that this was a custom job and not just an off-the-self product. Yes, the USB's are reusable, but if we REALLY wanted to save some bucks and the environment, how about just giving us free WiFi (it was $10 per day, I think) and let everyone download the stuff without further ado? Sometimes, the tree-hugging ecological outlook doesn't go beyond what looks good on the surface.

SIIM this year suffered from lower attendance than was seen last year, probably due to the economy, but also there might have been the fact that there isn't much to do in Charlotte after hours. I'd rather be in Toronto, personally. My friends at PACSAholic have done their usual impeccable analysis of the situation. The exhibit floor was packed with vendors, yet several, Fuji in particular, were notably absent. The floor was NOT packed with customers, as one might see at RSNA.

I only made it to two lectures, but I did sit through them from beginning to end. The first was the Keynote General Session, "The Design and Failure of Useful Things" by Henry Petroski, PhD. This was an interesting journey through the development of simple things like paper clips and straight pins up to the Kodak Carousel slide projectors. Sadly, Dr. Petroski quit at that point, and it was as if no one had told him that this was a meeting concerned with PACS. Oh, the places he could have gone with my concepts of Lego PACS and feature fatigue! The take-away message was that new products are developed to take care of deficiencies in the old products, and developments in new technology come from limitations in current technology. True enough.

The second session graced by my presence was a debate amongst five participants about "Who Actually Owns Imaging?" The session is summarized superbly on the STAT READ blog. The participants outlined the basic arguments for radiology-centric, IT-centric, and even clinical-department based ownership. The most compelling argument came from Paul Chang, M.D., developer of Stentor PACS, who called for enterprise IT ownership under the affinity matrix model. I would present it as a way to own the piece you understand best, with meetings periodically to make sure all is running well, as well as accountability for those involved. My interpretation is really back to what I've been saying all along, ownership has to be shared.

My visit to the exhibit floor was dedicated mainly to advanced visualization, otherwise known as 3D. I did stop in at the Amicas booth, which had a lot of new faces freshly imported from Emageon. I was very impressed with the grasp of the Amicas philosophy and product these new kids displayed. The big news beyond the actual release of Version 6 (Amicas PACS with Halo viewer) is their new vendor-neutral archive. It's rather hard to demo something like this, but their press-release states:

AMICAS Enterprise Content Manager is an enterprise clinical image management and information storage platform. Successfully deployed at many major hospitals and healthcare systems throughout the country, Enterprise Content Manager offers a vendor-neutral archive with proven scalability and interoperability capabilities that supports a broad collection of comprehensive electronic medical record (EMR) deployments.

"The demands on the infrastructure of imaging providers have never been more intense," said Paul Merrild, senior vice president of marketing and business development at AMICAS. "Given the impact of higher exam volumes and an explosion in the average size of imaging exams, the pressure to scale and support multiple imaging departments with a standards-based solution that is integrated with the provider's overall EMR strategy has never been more acute."

"AMICAS' vendor-neutral archive solution addresses these problems with a proven, standards-based infrastructure that supports multiple departmental PACS solutions - all while lowering the total cost of ownership of an imaging infrastructure over time," said Mr. Merrild.

Key features of the Enterprise Content Manager platform include:

Massive scalability with proven performance managing billions of exams throughout the country - representing petabytes of storage - via a high-availability architecture

Proven compatibility as a vendor-neutral solution with many major radiology and cardiology PACS vendors, which offers a lower total cost of ownership for the enterprise

A highly compliant, standards-based architecture with strict adherence to the IHE framework and protocols, which helps providers avoid costly data migrations as clinician preferences for departmental solutions change

Proven support for numerous provider EMR solutions at many institutions, including some of the largest healthcare systems in the country

I'm glad Amicas is on board with IHE. Amicas is not the first with a vendor-neutral archive, but it deserves serious consideration as we progress head-long into EHR-land.

Agfa has some interesting things to see. IMPAX Mobility, a rebranded version of EasyViz by Medical Insight of Copenhagen, is a limited viewer designed to be independent of platforms (no JAVA, .NET, etc.), utilizing only 1.5 MB of code, which utilizes server-side rendering for faster loading and response. It will integrate better to EHR's than big ol' IMPAX 6, and it has some fairly decent tools including limited 3D capability. You can demo EasyViz by signing up with Medical Insight. It does require an additional server attached to your PACS.

There is a second, even more limited viewer under development called Xero, which uses AJAX and is browser independent. It allows varying window/level and zoom within a browser window. This is geared toward Agfa's version of the vendor-neutral archive, which is in the works as well.

I started my 3D software hunt at the Agfa booth, a good place to start given the fact that if we actually get to buy anything this year, it will attach to or integrate with our IMPAX 6 installation. Right now, we have the Toshiba/Voxar software package loaded on each workstation. This would be replaced by Agfa's "Volume Viewing" module with full MPR, curved and oblique recons, VR, and so on. It works more or less like Voxar but with one little problem: The tools are embedded into the margins of the image. For example, there is a faint little hand icon that lives at the top of the image window; you click it to access the "pan" function. I didn't like this sort of thing when I used it on ScImage, and I don't like it here, either. Too much skirting all over the screen, you know. The PET/CT module has similar controls, and goes a step farther (behind, if you ask me) by having the blend function controlled by clicking and dragging over the word "Blend" at the bottom of the image. Uh....no. As of yet, one can only view one PET/CT at a time, without the ability to compare old and new. That's coming, I'm told. There will be auto-registration of any two modalities, not just PET and CT, with the adanced Fusion/Registration program.

As for cardiac work, Agfa sells the HeartLab product, and to do coronary work, HeartLab opens a window into.....the TeraRecon server, which you have to buy if you want to do coronaries.

Which takes us to TeraRecon. This is pretty much the 900 pound gorilla of the advanced imaging field. I won't even try to do it justice in this brief description, but suffice it to say that there is a considerable degree of automation, and that most functions are available over the network via thin client. The number-crunching is performed server-side. TeraRecon uses Dell 2900 and 2950 servers with a proprietary rendering board. Adequate bandwidth is needed for thin-client access, miminum of 3 MBPS, more optimally 5-6 MBPS for advanced functions. The product is sold by the number of simultaneous users. This can be deceptive. I predict that only a few rads at a time would access the system. However, once we get the surgeons and the like on board with this, we might have many more folks banging on the servers. It's a good inducement to go for the unlimited package, which is pretty pricey.

With TeraRecon's Aquarius iNtuition package, you get automated coronary evaluation and report generation, and automated virtual-colonoscopy processing as well. There is similarly mostly-automated reporting of stent evaluation, (EVAR) which supposedly can generate $300 reimbursement per case. The PET/CT tools are a little limited with manual registration of an old and new study required, although autoregistration is said to be in the works. Brain perfusion is available with pnumbra overlay. Tera has all the usual 3D bells and whistles, with automated table and bone removal, and relatively easy production of pretty 3D movies.

The other big player in the 3D sandbox is Vital Images. However, as I reported a long time ago, we owned several Vitreas, and when one broke, I couldn't get anyone from Vital to call me back. They found me on a cold call years later, and still didn't have much to say. Sorry, Vital, since you didn't have time for me, I didn't have time for you.

I have made a bit of fun of Visage Imaging in an old post, mainly because their old parent, Mercury Computing, bought a German company named SoHard, and I just couldn't let that go. Visage is now owned by Promedicus of Australia, by the way. Maybe we'll call Visage CS the 800 pound gorilla. It offers a significant part of the TeraRecon palatte, exceeding it in some areas, for a much lower cost.

Visage uses HP servers, with a Windows database server, and a Linux rendering server. Each rendering server can handle six concurrent users. Like TeraRecon and Voxar, the thin-client can be integrated to Impax 6. Toolbars can be configured and the whole thing runs on one or two monitors, as you wish. There are the features we have come to know and love, like one-button removal of ribcage and scan table. The degree of automation of a coronary exam is not quite that of TeraRecon, but it still works quickly and cleanly. There is no automatic stenosis detection, a deliberate move, as the Visage developers felt that too many errors were introduced by the detection algorithm. There will be automatic detection of the coronary tree in the next version, or so I'm told. Calcium scoring is semi-automatic; the program finds the calcium densities, and one then "lassos" the artery to select the proper vessel. Brain CT perfusion is automatic, just select an artery and vein, and off you go.

Visage does better with PET/CT than TeraRecon at the moment, and almost exceeds the performance of MIMVista, which is to me the gold-standard. As with MIM, Visage has the capability of auto-registering a new to an old PET/CT, as well as registering the PET component to the CT component. The one thing it doesn't do is automatically propogate the marked lesions from the old to the new exam, but there are provisions for doing this manually, and then giving the percent-difference in volume (and I think SUV as well).

Visage prides itself on using off-the-shelf equipment, particularly the latest in nVidea video cards. The service/upgrade program routinely replaces the video card with the latest and greatest.

Based on what I've seen, our choice between Visage and TeraRecon will be made primarily on funds. TeraRecon is a more powerful product, but Visage is right up there, with maybe 85% or so of Tera's functionality. Visage is considerably less expensive. I won't say how much cheaper, but "a lot" would be a reasonable description. If you have lots of cash to spend, go with TeraRecon, but if money matters, and it does for most of us these days, Visage is a very reasonable alternative.

We plan to bring both workstations in for a side-to-side comparison (shoot-out!) and I'll report on the results when that happens.

If there is something you saw at SIIM that I didn't mention (the vast majority of toys, sadly), please enter a comment with your take on it.

I'm not sure I'll make it to SIIM 2010 in Minneapolis, but look for me at RSNA this year. I have to get data to feed my audience somewhere!