This morning, I received the following comment on my "Good, Bad, And Ugly" post:
I just want you to know that your problems are not isolated. I have been watching an Impax 6.0 rollout since last summer at another institution and have not exactly like what I have seen. To me (an IT/business person) Impax 6.0 is a shiny new wrapper on the old Agfa PACS. Yes there is a new front end and yes there is the ability to tie into the system using webservices, but I am not sure any of that matters. Many of the radiologists do not like and want the old version back. To me AGFA really needed to hit a homerun with this version. Anything short would not be enough. I just don't think that they did.
Wow. This is what I have been saying for a while now. Agfa ported the viewer pretty much in toto from Impax 5.2. Some of my partners are showing some nostalgia for 5.2, but in the end, I don't think Agfa changed a lot overall. Sadly, they brought forward most of the bad as well as the good.
My whining may have had some effect, although our hospital's chief PACS administrator really wielded the heavy club for us, and I am forever grateful for her efforts since she finally realized just how much trouble we are having. I don't know all that was said to Agfa, but I can guess, and as a result we will be visited by some of my good friends from Waterloo this coming week. (If nothing else, they will enjoy the weather down here.) I'm very glad they are coming, and what I hear of the composition of the group, they will actually have the power to do something about some of our problems. Personally, I hope they have the ability to do a near-complete rewrite of the system, because that may well be necessary. Sitting here, on call on an otherwise lovely Saturday morning, pounding away at Impax 6.0, it is pitifully obvious that Agfa did NOT test this program in any sort of high-production environment with rads that were familiar with other products. Agfa has told me differenly, but I just can't believe it based on what came out the other end. (And that is a deliberate allusion.) Here is a partial "doo doo" list (hey, this is a family blog!) that I hope to see addressed:
- Crashes, crashes, crashes. We have found one or two things that will trigger a crash, but for the most part they happen randomly. I thought CRL and .NET were so very much more stable than JAVA.....
- As often as not, the wrong study among a series of new and old studies will be selected for on-stage status. More often than not, the wrong comparison will be selected. I spent 10 minutes with a clinician poring over a mass that had apparently grown rapidly, only to find that Impax 6.0 had brought up a much older comparison. Egg on my face, and on Agfa.
- Series bar...a very poor implementation indeed. The series bar holds thumbnails for a study that has several different series, such as an MRI. The bar has inconsistent behaviour, usually pops up over what you want to see, gives unintuitive control over distributing the series, etc. This needs a big revamp.
- Things like rulers and elipses may not stay active long enough to actually use them. There is nothing so frustrating as placing the measurement cursor over a nodule, clicking, and having the whole ruler revert back to the window/level tool or something before the measurement is placed.
- TOGGLING....Please tell me who approved of the "toggle" concept, and I'll make them listen to my poetry until they beg for mercy. This, of all the Agfa approaches, is the one that makes me tear my hair out. It leads to lots of other problems, probably including number 4 above, as well as constant re-deployment of Voxar 3D in an otherwise excellent integration. By toggling, I mean that if you click a button, it is still active until you unclick it. This leads to a ton of problems, trust me, and Agfa is, to my knowledge, the only company doing this. It needs to be ditched, and brought back to the way everyone else does things, that being only one tool at a time. If I click window/level, my tool should stay that way until I click something else. Actually, the window/level tool may have been the impetus for the design in the first place: Click the tool, then click in the image, then manipulate the window without another click until you like what you have. But why not just do the click-and-hold thing that EVERYBODY else does? Yes, getting rid of this is a major rewrite, but you guys really should have thought of that before inflicting it on us.
- Window Toggling is another major problem. The color or outboard monitor shows a state-dependant display, geared to which exam is selected in the viewer. Even after several months, it is still disorienting, and it can lead to problems like those in number 2 above.
- If two studies or exams are displayed, tools will only work on the one in the leftmost monitor. Applying a tool such as window/level to a study or series may cause it to shift to another monitor.
- The screen is rather slow to refresh or repaint when moving to a new study or series. If reading from a worklist, and for some reason, you didn't read the first study, that first study flashes briefly on the screen before the next study in line loads.
- The "Search" function is a complete disaster. Either a simple search (which I have never seen) or an "Advanced Search" is available, but NOT both. Really shortsighted move, boys and girls. What do I search for about 95% of the time? A patient's name, and to get to that on the advanced search, I have to scroll half-way down a list of two dozen possibilities, then set up a search. This was probably the brainchild of a brilliant engineer, who had absolutely no idea of how we actually do things. It is incredibly powerful, but incredibly frustrating the vast majority of the time. Please just give us the simple search in addition to the advanced extravaganza, not instead of.
- There is no easy way to get back to a study you have bombed out of, say due to a crash, unless you happen to remember the patient's name. Even 5.2 had the old bookmark that held the name of the last viewed patient.
- The concept of clone windows as opposed to being able to drag any series to any viewport is a pain. How about doing what the other folks do?
- Hanging protocols are still unusable by the average doc. There has been an improvement in their set up, in that one no longer needs to understand machine language, but the new and improved drop-down interface is still going to be the domain of programmers. Sorry, we need something we can actually operate at the user level.
- Table position linking works fairly well, although I much prefer the Amicas method wherein you press one button and that's it. Sadly, Impax 6.0 has a very nasty tendancy to, ahem, toggle back and forth between one-on-one and multiple windows at just the wrong time, such as when I'm trying to link two windows. There seems to be some sort of a mouse click timer issue here. Please fix it! I've almost smashed my mouse five times in disgust over this one just this morning.
- Hot-key programming seems to be retained only intermittently. I'm wondering if there is a problem transferring the settings between our THREE application servers.
- "Sticky" window settings, column resizing, and the like, is not always, well, "sticky". Perhaps this relates to 14 above. Possibly related is the fact that sometimes window/level changes will apply to the whole series, and sometimes not, requiring the reactivation of the series-wide control. Why doesn't it just stay put!
- As noted in previous posts, we have been thrown for a loop by the fact that things that were supposed to be individualized are not in the production model. Having items such as window/level presets, precanned comments, etc. set up as global for a user-class, i.e., radiologists, has led our own PACS people to take the control away from us, leading to more problems. In particular, the button to add precans still exists, but because we can't use it, it bombs the program. Give me back my individuality!
- The measurement tool works differently than any other in the business, mainly because it places the measurement right in the middle of the line you have just drawn. Now there is a little handle to grab to move the number out of the way, but if you are measuring a small item, that handle is impossible to access, and the number is left to obscure the lesion. How about throwing that number off to the side? Even ScImage does that!
- And lest I forget, there is no workable spine-labelling tool.
This is only a partial list...my partners are helping me put together a much longer and more extensive tome. I have high hopes for the impending Agfa/Mitra visit. I just wish they had come by to see how an average group operates BEFORE they wrote Impax 6.0. But they didn't. So, we are in a jam. We have a system that is no better, and in many ways worse than the old one it replaces. The only real improvement is web-access, and that is indeed a step up from good old Web1000. Still, we are left with few choices. We can suffer through these problems, waiting for Agfa to fix them at their leisure, we can demand to revert back to Impax 5.2, or we can insist on a forklift replacement and go with another PACS. At this point, none of those are really acceptable. I can only hope that the Agfa/Mitra team is committed and empowered to fix this thing, although I'm not too optimistic given the fact that it took over three years of development to get it to this point! Still, if things are not going to be fixed in a very timely manner, option three starts to look better and better.
In the meantime, patience on the part of prospective buyers is still advised.
My friend quoted above sent the following comment about the above post:
Hi Dalai,I will take issue with your first point about .net/CLR stability. So Agfa picked a good infrastructure to use. But just because they did does not excuse them from writing good code, which in my estimation the code quality is probably poor. There are many rock solid .net applications out there. Impax just isn't one of them.You mention testing and I really agree. But it isn't just an Agfa problem. The development practices of PACS software run a gamut from good to bad, but most lean towards bad. I have observed issues that should have been caught in testing. If there is an extensive test plan in place and these issues are still being had then something needs to change internally.Let us know what happens.
Rather a harsh indictment of the programmers, but.... I certainly agree about the testing aspect. Note to vendors: Get a wide range of users to beta-test your latest and greatest. You might learn a great deal. Note to buyers: Do whatever you can to test your intended PACS product in a production environment. The more you live with it, the more you will come to know its good and bad points. Simple advice that no one follows until it's too late.