A few weeks ago, Agfa had determined that over a period of time, server resources accumulated to the point that one of the servers would freeze up. Since then, they had put a process in place to release system resources, which has greatly decreased the frequency in which the client failed.Unfortunately, in order for us to determine the root cause of the problem, Microsoft has requested that we gather a “Crash Dump” log file off the server. In order for us to get this file, we need to disable the process written by Agfa and wait for the system to degredate. This may cause the clients to begin crashing with more frequency, however, I want to inform you that this is in efforts to determine the real nature of the issue.
When E.F. Hutton talks, people listen. When Dalai whines, people cover their ears. Actually, that isn’t always the case. We were honored down here at our average hospital in our average town in the South to host a smaller Agfa entourage which included Bob C., Agfa’s Senior Marketing Manager. (No, he does not sell old people.) I first met Bob at RSNA 2005, where he gave a masterful demo of Impax 6.0. You might recall my observation at the time that my visit to the booth was somewhat surreal. I had written about Impax 6.0 on my blog, they had all read it, and it was pretty clear the Agfa folks didn’t really appreciate my literary talents. Oh well.
But fast forward to this week. Everything got off on a much better foot, and I think perhaps there might have been a smidgen of mutual respect going on. Of course, it helps that Impax 6.0 really has been showing its behind, and everyone realizes that I’m not totally full of it. I was able to spend about an hour with the Agfa folks, before my anxiety about the growing worklist made me bid them adieu. This was a very productive meeting, I think, at least for the long-term. Bob was able to see how Impax 6.0 functions (or doesn’t) while in my hands, and I think he saw some things in ways he hadn’t before. Take, for example, the concept of tool-toggling, which I harp on constantly. Bob had not considered that keeping a tool active until deactivated and then falling back on a default tool might not be intuitive. Agfa is the only company with this approach, and I for one don’t like it. The idea was apparently to limit the “mouse milage”, i.e., to keep one from having to move all over the screen to activate another tool, but in the end, I don’t think it accomplishes this. As a corollary, even the window/level tool works differently than with any other product…left-click to activate, move the mouse to change window and level, and left-click to deactivate and go back to the scrolling tool. I’ll take the way everyone else does it, thank you. Bob seemed surprised about that. We went on to discuss where the series bar should go (we agree it should be fixed at the top of the screen), and how to display old reports (I gave him my very simple idea of putting the date of the old report in big red numbers and letters at the top). We talked about a new little peculiarity Impax is manifesting, flickering upon viewing a newly-loaded CT, apparently due to slow caching, although it happens inside the enterprise as well as outside. Bob noted that there were several improvements that I missed at RSNA whilst frolicking amongst the 64-slice and beyond CT’s. They will shortly have a spine-labelling program that in Bob’s words is better than Amicas’ (although from his description, I think it will be quite similar).
The meeting was much more cordial and informal than the blog-inspired stand-off of a year ago, and to be honest, it was probably more comfortable than the talks we had with the developers a few weeks back. In the former scenario, we had the distinct impression that we were calling someone’s baby ugly. Well, I’m sorry, but for us, it is. Perhaps as head of marketing, and not the “parent” of the product, Bob can be a little more conciliatory toward our views. Case in point is hanging protocols. In Impax 5.2, it took a computer genius to apply them. In 6.0, it takes at least a PACS guru, although I probably could get one going in an hour or two if left uninterrupted (yeah, right). The developers were absolutely incredulous that I might want to make my own, but Bob expressed understanding as to why I might have that desire.
I think I sent the Agfa folks back to Canada on good terms, and I’ll tell you, I am very impressed that they made the trip. This is the way to build a PACS, folks, although Agfa is about three years late in the process as far as I’m concerned. The keys are simple: listen, watch, and learn. I’ll bet Agfa knows more about how real, live, average radiologists work now than they did before coming out here to the boonies. You simply cannot design a product based on a focus group of 5 or 10 or 20 rads. You need to see how a broad swath of them use a product, and go from there. Now, I violently disagree with the philosophy of giving everybody every bell and whistle they request. With that “please everyone at the expense of everyone else” attitude, we end up with the products we have now, bloat-ware with a thousand options that in the end are rarely used. Can you say “Feature Fatigue”? It is unwise in the extreme to add a dingus to please 1% of the buyers. As I have told Amicas a zillion times, keep it simple, or you’ll be no better than those other guys. I think Bob might actually agree with me deep down, based on some of his comments, and he alluded to a “light” version of Impax 6.0 that I might have found more to my liking. At least he can see this issue from my point of view. But the important thing again is that he came to my place to see how I and my partners do things. This was incredibly important to us, and I hope to him. Every PACS manufacturer should consider this approach. Maybe I should charge admission.
I am about ready to lift my “Wait” designation on Impax 6.0. The workstation crashes have tapered down to only once or twice a day, much improved from every 10 minutes or so. From the hospital’s PACS administrator comes the explanation:
So, maybe the problem will be solved for good. I do recall that there were similar problems in the beta install at Sparrow Hospital in Michigan, caused by some Microsoft or Windows problem that impacted Impax. Is history repeating itself? There are a lot of nay-sayers out there that aren’t thrilled with any strong dependency on Microsoft, and this sort of thing does make you wonder. While I’m not that big a fan of the GE AW, it does run on Linux, and I have never seen it crash. Well, maybe once, but it is pretty darn stable. I think Brit Systems with their Linux PACS product may have it right after all.
Anyway, once the crashes in Impax 6 are fixed, I will change my warning from “Wait” to “Caveat Emptor”. If you are considering this product, you or your rads are behooved to demo it to death. Get as much hands-on time as you want, and don’t take “no” for an answer on that request. The only way you can tell if you will be able to live with Impax 6’s peculiarities is to put it through its paces (or more accurately, have Impax put you through your paces). That is good advice for any PACS system, and indeed just about any high-level purchase. I’ll let you know if this Microsoft issue turns out to be the real culprit.
And again, Bob, thanks for coming down. It means a great deal to us that you were willing to take the time to listen.