Saturday, January 29, 2005

Amicas vs. Fuji 1/05

Posted to AuntMinnie 1/05...

I now have Amicas LightBeam PACS at two of the hospitals I cover. The other, larger hospital is upgrading Agfa Impax 3.5 to 4.5. A large oncology clinic we staff uses GE Centricity 2.0. We looked extensively at several systems. I will spare everyone the gory details, and stick to the question at hand. Maybe someday I will team up with Mike Cannavo and write up the whole search saga. It might make a good mystery novel....
I am very pleased with Amicas, and I would buy it again in a minute. All the rest is commentary, as they say. My partners now pay them the highest complement: when referring to PACS in general, they usually say "Amicaspacs", all in one word. In brief, the AmicasRealTime Worklist is incredible, and that is what sold me on them at SCAR in Boston in 2003. You see what you need to see at a glance; you know what needs to be read, what has been read, and the status of various things right there in a clear, color-coded format. The list can be customized any way you want it, even to the point of having a different set of lists for each user. The viewer itself is very straightforward and intuitive. The few major tools you need about 90% of the time (scroll, window/level, pan, zoom) cycle by a quick right mouse click, and all other tools are either on the menu bar of available through a right-click-and-hold pop-up menu. There is an embedded subset of the Voxar 3D tools, and there can be integration with the full Voxar 3D program or alternatively with a TeraRecon server.
As near as I can tell, the guts of the system are similar to Fuji Synapse, using Windows Server, for better or worse. Both are completely web-based, i.e., each image has its own URL. Both deploy software over the internet/intranet/LAN, i.e., the "thin client" approach. One's opinion on an interface is just that, an opinion. Some like Fuji's client; I am among those who do not. They use a folder analogue instead of a worklist; that is OK as far as it goes, but every time I tried it, I ended up with folders all over my desktop. Probably just user error, and I have been flamed repeatedly for disliking their system. I have been told by a senior exec at Fuji (who since left and went to another similar company) that I was far from alone in my feelings about their interface. In our PACS search, we actually hit two Fuji demo sites, one in Austin, Texas, and one in Virginia. The Austin site has a spectacular communications network, set up mainly by the experts within that particular radiology group and Time Warner Cable, and NOT by Fuji, again as nearly as I could tell at the time. Studies were delivered fully within 3 seconds of the request. This came at a tremendous cost in communications; thousands of dollars per site per month. One thing becomes clear when you talk with Synapse users: They like the product, but they seem to have to do most of the maintenance on their own. A common theme is that Fuji has not yet mastered the sales or the service on an otherwise good product.
Amicas has not yet reached the level of perfection, either, lest I leave you with that impression. Our install has taken longer to come up to 110% functionality than I would have guessed. Some of this is due to the fact that we have only one PACS administrator for two good-sized hospitals, and his helpers from IT have a thousand other things to do at any one time. Still, I am overall satisfied with what we have to date.
Agfa is a whole 'nother story. I am willing to slog through Impax 4.5 in hopes of eventually having 6.0 installed sometime in late 2005. (I urged that hospital to put some onerous penalties in the contract should 6.0 not appear, but I don't know if they did so...)
Bottom line: Amicas is a good company with a good product. They do have a significant number of "real" PACS systems installed in large hospitals. I recommend them highly. For what that's worth, anyway....

My SCAR 2003 Review

Posted on AuntMinnie.com June, 2003....

Many of us are in the PACS market, either to replace an existing system, or to dive in for the the first time. Hopefully, many of us went to the SCAR meeting, which provided a great forum for discussion and comparison of the various products out there. I want to compare observations. The purchase of a full PACS system represents a multimillion dollar investment, and we all want to find the best possible solution. Therefore, I shall post my own observations and I BEG you all to do the same.
General Observations:
1. Everyone is headed to "web based" architecture. This means for most that each image has its own URL. Some are already there, some are getting there, some are planning to be there someday.
2. MPR/Volume rendering on the PACS workstation is the up and coming thing. Most vendors use either some limited proprietary method, or bundle the Voxar 3D product in some form. See below.
3. The three monitor set-up is just about universal now. This entails two high res monochrome monitors (generally 3-5MP) with a small color monitor for the worklist and maybe some 3D stuff. EVERYONE had flat panels, usually Barco's for the high res. The vast majority used Dell workstations.

Company Specific (far from complete....):


AGFA: The grandfather of PACS..seems to have lost a tremendous share of the US market. Their client is very good. Sadly, their underlying architecture seems to be similar to the original (distributed databases, routing, etc.). The good: new product in the works... The bad: no one knows when it's coming.

Algotec: A good front end as it stands. The completely new client was not ready in time for SCAR. It is said to offer much 3D capability. Architecture is a little unclear; rep emphasized the "modular" capability of the system...ie, when expanding or upgrading, one box can be reconfigured or redesignated or reloaded to do the work of another box, thus preserving your investment.

Amicas: To me, this was the "sleeper" of the whole show. My personal opinion is that they have the best combination of client, worklist, 3D integration (with the Voxar engine), and web-based architecture. I would have placed my order then and there, but they do not have any installs as large as mine would be, and I have learned not to volunteer for beta testing. Frankly, I'm hoping they can convince me they CAN handle such a large project. They say that what they were showing was what they were shipping NOW, a nice change.
Fuji: Probably has the strongest web-based design. EMC reps felt that Fuji was best able to use their storage products for greatest speed and efficiency of all the PACS vendors. Again, a personal opinion, but I HATE their front end. The web/folder model is cute, but I find it much harder to work with than most of the others. Voxar 3d is thrown in as a free standing program, integration is promised, but wasn't demonstrated. They too say that what they show today is what they are shipping today. In speaking with other users, the general opinion is that Fuji delivers what is in the contract, AND NO MORE, so negotiate wisely. They are very price conscious, and make it clear that you will pay for what you get. Hopefully, you get what you pay for, too.

GE: Ah, the General! What can you say? They will probably supply more systems than any other company simply because they are GE. Centricity has a very usable front end, better if you have their VERY expensive RIS to go with it. Looks like their hanging protocols are great once you go through a rather lengthy set up process. I had a hard time pinning them down as to what ships when. Apparently, Centricity 2.0 will NOT SHIP UNTIL 12/03. They are flailing around with umpteen 3D solutions... I think the TerraRecon server box option will be available to ship with Centricity 2.0. They also showed their Advantage Workstation (AW) in a similar configuration (ANOTHER server box plugged into the network with "thin client" operation from the PACS workstation) which actually worked well, but won't be available for you and me until 12/04 (maybe). The 3D module of Centricity itself is a joke. I had heard something about a software port of AW to the Centricity platform, but all they had was an old AW package that didn't work too well. Underlying architecture is NOT web-based in the individual URL sense. They do however use centralized database structure, though it is two-tiered. Still should be functional and expandible. They do use UNIX in the central components...may be good or may be bad depending on your point of view. Personally, I still have some love for UNIX; I can't get it to crash anywhere near as often as I can Windows of any flavor. The General has acquired a reputation for running roughshod over us users about price. One big rumor abounds concerning a big academic install where GE was yanked out in the middle of the process and Siemens brought in. I had the opportunity to talk to one of the principals, and the truth apparently is that GE upped its price by several million and the institution decided to go with a cheaper alternative. I don't know if the institution changed its requirements or some such thing in the middle of the negotiations. I am anxiously awaiting comments from my GE reps....

Philips: Fairly good front end, most of the bells and whistles are included. I don't think they are web-based. Most of their main components appear to be from Sectra.

Siemens: Good client, but elderly architecture due for complete replacement in18-24 months.
Obviously, I didn't get to every last vendor. Please, please, please everyone, fill in the gaps and tell us what YOU think!

In the beginning.....

This was the post on AuntMinnie.com that started a war with ScImage and an Internet presence:


In a word, HELP! Our department was one of the first in the US to go filmless, umpteen years ago, using AGFA. Today, we are one of the last two departments in the WORLD still using AFGA 3.5. Our hospital didn't like the idea of a $2 million upgrade to AGFA 4.5. In the last 60 days, we have brought a 16-slice CT on-line at our main site, with 2 more to come. We also began reading 16-slice CT studies from an outside clinic. Needless to say, AGFA 3.5 chokes on these scans. It cannot handle the vast number of images in a timely manner. In addition, comparing a new scan to an old study with different slice thickness is an exercise in agony, as AGFA will not track the z-axis. We have been using a demonstation system from a small West Coast company to read the studies from the new scanners. (E-mail me if you really want to know which company....) In truth, we are beta-testing the system. There are numerous bugs, though to their credit the company will fix them within a day or two of their report. The system uses a lot of Java programming, and probably a lot of simple HTML as well for a web-based approach. The patient selection page is very cumbersome to use, but they have written a shell program to buffer this, and it is an improvement. The standard reading module is acceptable, with the usual buttons, and the obligatory right-click menu. The real selling point of this little system is its 3D module. We use the MPR portion of this component to read our 16-slice CT's. The program has adjustable slice/slab thickness for the MPR's, and the images appear to be MIPs...I'm not sure if they are really MIPped or if this is a perception from the slice thickness. One can "swim" through the MPRS as fast as you can move the mouse; the recon appears to be VERY rapid on a 3 GHz P4. The MPR module is still buggy, and is not integrated with the viewer. I can find a lung nodule fairly easily with the MPR, but then I have to wade through the source images to find the slice number to report for followup. The MPR module also has a backwards approach to triangulation; clicking on the lesion does NOT localize it on the other planes, but simply enables the plane you have clicked to scroll. I am one of the radiologists who has to use whatever equipment we buy for the next Gawd-knows-how-many years. I have some computer training (BSEE), and I see lots of bugs, shortcuts, and other danger signs in the small company's product. Not to air dirty laundry, but I find myself at odds with my colleagues about this. They love the (flawed) MPR, and they perceive the system to be cheaper than the big name stuff. They are impressed by the fact that the software gets rewritten for us all the time. (To me, this is like buying a Mercedes with a new, untested engine. Sure, the dealer will tweak it and redesign it and replace it when it misbehaves, but is this a way to function day in and day out?) So, my Forum friends.... 1. Is there a reliable system out there that combines all the necessary PACS elements with an integrated 3D/MPR program, AND z-axis matching to allow comparing old and new CT's? Might it possbly be reasonably priced? 2. How hesitant (or adamant) should I be about the small company product? Has anyone had good luck with this scenario? What happens if Mr. Small gets bought out by GE? Many Thanks!!!