Friday, February 18, 2005

Dalai likes the Amicas Real Time Worklist! Posted by Hello

Questions from a Like-Minded Rad

An email "conversation" about 3D, etc....

Hi Sam,
My name is A...... I'm an interventional radiologist and the designated PACS maven for our group which practices in a 400 bed community hospital in the greater ...... area. I've enjoyed and/or learned a bunch from many of your posts over the last couple ofyears and would really like to talk to you about a variety of topics including 3d, mdct, reporting dication and general workflow. I've spent the last three years reengineering all our departments processes during a Stentor PACS implimentation. Whilethe major stuff is done, there is a ton of tweaking left to do and I'd really like to just get some ofyour thoughts. Here are a few questions. If emailing takes too much time, I'd be happy to arrange a mutually convenient time to chat on the phone.

1. Does your Auntminnie "handle" have any religeousconnotation? I've been taking a Tibetan Buddhistmeditation class and just had to ask.

2. How much post processing beyond MPR do you thinkwe will be doing at our PACS workstations with MDCT? Should I make sure the techs stay real competent at all the 3D stuff or is it going to get so complicated (think coronaries) or fast (all automated) that we will do it? What are you using now?

That's a start. ....Hope this isn't too presumptious. Thanks, A....


Glad to talk with some like minded folks!

Actually, I am Jewish, quite reform, really. The Dalai Lama thing came about as an attempt to keep a vendor (ScImage) off my tail. I wanted a nickname that no one would ever associate with me, and the Dalai came to mind. I hope I am not showing any disrespect in this. Sadly, ScImage figured it out anyway. That's a whole 'nother topic.

I do Nuclear Med as a subspecialty; I'm one of the few who did a NM fellowship after DR. It is heavily computer oriented, and I have always enjoyed it. I'm an electrical engineer by training, although I never practiced as such.

I took over the PACs "guru" position in my group when the former guru got fed up with Columbia and went to Florida. He and I had a number of differences of opinion on how to do things. For example, he loved (and still uses to this day) ScImage, because they have one 3D module (Netra) that lets you view a whole volume with the sweep of a mouse. Basically it uses thick slap MIP reconstruction, and somehow loads it such that there is fast screen refresh. Sadly, the programming this one good module comes wrapped in is primitive, and it takes me longer to set up a study to be read than to actually read it. Bad sign. He insists that one must perform isometric imaging (thin enough slices that the voxels of a volume have all three dimensions equal), and then you have to visualize the new and old studies with a dual MPR program. Right now, ScImage and Philips are really the only ones that do this overtly. We have a GE AW (which I HATE!) but it can be conned into doing this as well. For what it's worth, I still do my comparisons with the good old axial images, though matched by table positions.

For 95% of what I do, the onboard MPR Amicas provides (I call it baby-Voxar) is adequate. Beyond this, I'll use the GE AW at the hospital that has that (If I really have to...) or full Voxar at the other. The other hospital has Siemens InSpace, and I really like that too, though I find I just don't make myself use it enough. (The GE hospital is in the process of changing from Agfa 3.5 to 4.5, and Voxar-based 3D tools are included as well as full Voxar...I'm not sure of how many "seat" licenses we will have...)

As to your question....We really haven't yet figured out the best workflow. My primitive thought is that still much can be handled with simple MPR, and I end up doing this on at least 50% of the CT's I read. My former partner says that you have to do the isometric voxel thing on EVERY SINGLE CASE, or you'll miss things, but I'm not convinced. I also must wonder if we are doing our patients a favor by picking up 2mm lung "nodules" that we end up following for years. Should the techs or the rads do the higher-level processing? That probably depends on your volume, your comfort level with the technology, the expectations of your referring docs, and a number of other factors. On more routine studies like angios and runoffs, where there is an easily-defined set of processed images, it certainly makes sense to have the techs take care of it. For the more esoteric stuff, it is probably wise for the rad and the tech to work together, at least until the tech is comfortable with it and the rad is comfortable with the tech's abilities on that particular study. All that being said, I find it incredibly valuable to be able to jump to full Voxar 3D and do a volume rendering and segmentation or whatever on those occasions when doing so will actually answer a question. It makes me wonder if I am asking the right question often enough, though!

That's a preliminary answer. I will attempt to hit SCAR this year and see what the rest of the world is doing. I'm thinking that no one has really found the perfect solution as yet. As with the adoption of PACS itself, it is a process everyone has to go through and mold (and be molded by) to reach the proper end. We are certainly not there yet!

I would really appreciate knowing your approach to all this!

Thanks for contacting me....


Thursday, February 03, 2005

PACS Selection Committee 2003, part I

I tried to convince the PACS Selection Committee to reconsider Amicas after they had eliminated all but GE, Fuji, and Agfa.....

To the Committee:

First, I would like to thank ........ for clarifying this issue. The minutes of a committee within a public entity may be subject to scrutiny, especially with a project of this magnitude, and they must reflect events exactly as they occurred.

Secondly, our experience with ScImage has taught me that one person should not fiat the purchase of a system as complex as PACS. The expertise of many has to be brought to bear. I was placed on this committee because of my Radiology experience, as well as my training in computers and electronics, and in addition, I have gathered a fair amount of information about PACS over the past several months. Even so, I was not allowed to review much of the data that came into the ........, and my opinion was not sought. To function as a committee, all members require full access to information, and full participation in the decision making process. I intend to participate in this decision; I would not try to make it myself even if that were possible. To that end, I will try to condense what I have learned into the next few paragraphs.

As I stated in the meeting of November 24, 2003, I have had the opportunity to test the products of all seven original vendors at the Society of Computer Applications in Radiology (SCAR) meeting earlier this year. I have been in touch with numerous friends, colleagues, and references concerning several of the products, both by phone and email as well as I have been on site visits sponsored by GE and Fuji. Although they were not made directly available to me, I have reviewed the Cap Gemini/Ernst & Young report as well as the replies to the RFI from Philips, Amicas, and GE. I have also reviewed recent KLAS reports and product comparisons from Reilly Communications, who publish Imaging Technology News/MEEN.

The Cap Gemini report narrowed the field down to Agfa, Amicas, Cerner, Fuji, GE, and Siemens, with Philips apparently being added later. (Most of us were not aware that Philips was already present within our hospital as a Cardiology PACS.) When I reviewed their report, much looked familiar; most of the information appeared to be either a recapitulation of the material we gave them during interviews, or alternatively a summary of facts from the vendors’ web sites.

As above, I have examined only three of the replies to the RFI. My read was significantly different, apparently, than that of others. Philips, GE, and Amicas all appeared to be able to fulfill the requirements. GE sent a veritable encyclopedia, fleshed out by dozens of pages of Xeroxed standard information. Many answers were in the form of long duplicate paragraphs that were copied and pasted into several locations. Amicas answered most questions too briefly, basically saying “yes” when they could perform a function. Philips’ response appeared to have been prepared the most carefully of the three.
The status of the individual companies in the PACS market may best be derived from the KLAS report; the most recent is summarized in these graphs, the first from 7/2003, and the second and third from 4/2003:

Klas Vendor Ratings Posted by Hello

PACS Selection Committee 2003, Part II

Klas 2003 Posted by Hello
A comparison of market share, etc, may be found at the Reilly Communication site: Several trends are apparent, but the most critical is that of web-based technology. This is where all vendors are headed, and all will admit this fact. There are only three companies among the major players that are at this point NOW: Amicas, Fuji, and Stentor. Most of the other vendors have told me that they will implement a web-based architecture sometime in the future, though only Siemens gives the timetable of 18 months or so to do this. Count on a significant hardware change out within the server to accomplish the move. All vendors now use PC/Windows for their workstations, but some still use Sun/Unix computers for database/server applications. By individual vendor: AGFA: Obviously, we have history with them, and they have done a superb job of prolonging the capacity of equipment well beyond the end of its useful life. We currently are one of three remaining IMPAX installs in the world. The system cannot handle data from our GE LightSpeed 16 CT scanner, necessitating an emergency interim solution, which likely will be provided by Amicas. Two years after our system was installed, AGFA changed their workstation platforms to Windows/PC from UNIX. An upgrade would have cost $2M. However, the upgraded system might have handled the 16 slice data. Agfa was once the market leader; in fact it dominated the market for many years. Its market share has slipped considerably, especially in the US, with far fewer new installations than GE or Fuji, the new leaders. As above, Agfa does plan to convert to a web-based architecture, though I was not told when this would occur. There would have to be a complete change of the server hardware to accomplish this. I tested IMPAX 4.5, and the client is actually quite good from my standpoint. It uses a limited form of Voxar 3D. Several AGFA sites are now using Amicas for web distribution. AMICAS: Let me state in no uncertain terms that I do NOT think this is the perfect system by any stretch of the imagination. However, this is a very versatile product from a very innovative company, and deserves further consideration. Amicas was purchased in the last week by VitalWorks, a fairly large company dealing in web-based RIS systems among other products. The purchase price was a fairly substantial $30M. VitalWorks has pledged to maintain Amicas as a wholly-owned but independent subsidiary. Rather than cloud the future of Amicas, I feel this acquisition ensures Amicas’ presence in the marketplace for the foreseeable future. The product is totally web-based; it IS a web-server at its essence, and it uses a non-proprietary form of DICOM which allows far easier migration than with other vendors. The LightBeam viewing client is one of the more user friendly out there, and it includes a worklist function that is second to none. There are monitoring and administrative functions for the PACS administrators and remote monitoring from Amicas. This company deals with software only, and hardware must be maintained by the hardware vendor (Dell, IBM, Compaq, etc.). Sites I have contacted have had little problem with this. (I would personally want Amicas to interface with the vendors such as Dell, to initiate any service call.) They do not have a Cardiology solution of their own, but can accommodate several third party programs. There is a limited form of Voxar 3D, with new memory management techniques to allow very rapid deployment of the 3D module (or optionally the full Voxar 3D program) from the viewer. Cerner: This system does one thing quite well: it interfaces with their Cerner RIS. Otherwise, there is little in its favor. Apparently they claim 20 installs; I have only been able to confirm about 5, and rumor has it that they never even mention one very troubled site (Detroit?)
Fuji: Synapse was the first web-based PACS system. Their architecture appears very solid. The viewing client is rather esoteric; many like it, some don’t. I have spoken to one of the senior Fuji VP’s in charge of the Synapse project, and a new version of the front end is in development. I have been less than impressed with the regional Fuji representatives. They generally do not know the capability of their system, and have to call back to the head office to get the answer to any question. At SCAR, it was made very clear that Fuji would monitor usage and charge accordingly. Fuji users I have met have been satisfied overall, but note that “you will get what’s in your contract, and absolutely nothing else.” Our site visit to Austin was billed as an example of Fuji interconnecting several hospitals. The system did seem to work well, although there had been a 5 hour outage the day before that no one could explain. The interconnection, and really the main functionality, had NOT been designed by Fuji, even though they implied otherwise, but rather had been set up by the group in Austin and Time Warner Cable. There is limited integration of Voxar 3D; clicking a button launches it but that’s the extent of it. They may be replacing it with their own 3D program. GE: The largest company in the business. Centricity version 2.0 has been touted for over a year, and might actually ship in December. The client is very well done, except there is not a set 3D solution. GE actually started selling Voxar 3D as an option after I got the two companies together. The underlying architecture is very complex, and I’m not even sure GE knows how it works. I spent about an hour trying to get their PACS people to tell me if it was web-based; finally, we conclude it is not, but it is web-enabled. This means there is internet access into the main database, provided by yet another box attached to the system, but the system is not a web-server in and of itself. They will change over eventually, and a new server and other associated hardware will be needed. Centricity is acknowledged to be the most expensive PACS system of all. A system was to be placed at Emory, but after GE’s price went up by several $M, Siemens was brought in instead. (GE tells me that Emory asked for a much expanded system accounting for the increase.) Installs at (local sites) have had numerous startup problems, according to the docs and techs. Radiologists at the hospital are now more or less satisfied with the operation of the system. Philips: Their product is made by Sectra of Sweden. Philips has more world-wide installs than any other company. The user client is quite user friendly, though not really particularly distinguished. It is modular, and will accommodate third party software for 3D, Cardiology, etc. Service and uptime are extremely good by report. Architecture is of the old style, distributed database. As with AGFA, GE, and Siemens, there will be a web-based replacement, though no one at Philips could tell me when this will occur. Siemens: Sienet uses a very unique client, though it is a match to their other eSoft components. (CT’s, gamma cameras, etc.) The architecture is of the old model, and as above a replacement is planned for about 18 months from now. Of all the vendors on this list, Siemens seems to get the most complaints. Summary: Forgive the long discourse, but this is a brief distillation of what I have learned over the past several months. In short, there is yet no perfect system. Several companies were eliminated by the initial Cap Gemini survey that likely should have stayed in the mix, including DR, McKesson/ALI, and Stentor. Even working within the Cap Gemini suggestions, the review committee’s recommended cut keeps AGFA, apparently for no reason other than the fact that we are familiar with it, and removes Philips and Amicas, which are still very formidable players. I understand the limitations of time involved in site visits, but I feel any extra effort required will be rewarded with a more optimal decision. Therefore, I strongly disagree with the current recommendation. The roll-out date has already been pushed from February, 2004 to May, 2004. I recommend we revisit the choice of vendors, and take the time necessary to utilize all information available. There is much expertise on this committee which has yet to be tapped.