Thursday, November 16, 2006

"Take One For The Gang!"
...The Agfa Development Team Comes To Town

My previous posts about Impax 6.0 have been the source of much conversation, especially in Waterloo, Ontario, where Impax 6.0 was developed. Due to various complaints, including those posted on here on the blog, and also many from my partners filtered through our IT folks, a high-level group of Agfa developers did indeed come to visit from the far North. They spent two days with me and with my partners, and I think everyone would consider the time well spent.

I think the poor guys were expecting something like Osama bin Dalai to greet them, not the adorable little fuzzball that they found here. I'm pretty benign in person, at least most of the time. Apparently, the team was given a send-off suitable for a departure to Baghdad. "Wear your Agfa condom, and thanks for taking one for the gang!" Nice to know that I'm so feared everywhere but my own backyard, where I'm either ignored, despised, or laughed at. When I asked why they thought I was such an SOB, my very own hospital's PACS administrator piped up, "It's because of your blog!!!" Thanks. I never would have guessed that. But, the visit was really productive nonetheless.

Impax 6.0 must have realized that it had more to lose if it made ME look bad than if it made its parents look bad, and like any petulant child, it misbehaved quite badly the moment the Agfa folks walked in the room. My workstation displayed the dreaded error and crashed three times within the first 5 minutes of their visit. I swear I didn't monkey with anything to cause that. Really. To ice the cake, the whole system went down for about 30 minutes later that afternoon, which I have heard was due to someone putting on a Microsoft patch without checking with Agfa first. Easily fixed. But this all set the tone for the visit. As of yet, we don't know why the workstations are crashing.

I have openly wondered just how Agfa tested this product, and if there were any real, live radiologists involved. I asked this question again, and again, the answer is yes, and in fact there were about 20 of them. I could not really drill down to whether or not these were Agfa users exclusively, but I rather think they were. That might explain to some degree why the viewer portion is a direct port of that from 5.2, which was a tweaked version from 4.5, which was a tweaked version from 4.1. The differentiation was made, however, between testing like they did, and in-the-field real-life observation, like they were doing with us. The former got them to where we are today, and the latter will get them to fix where we are today, and go forward to a bigger and better system. Well, a better system, anyway. This is probably the take-home lesson for Agfa, and really for ALL PACS companies. There needs to be a great deal more effort toward involving practicing, PACS-savvy radiologists at every step of development. Granted, they won't agree on everything, but they will ultimately hash out a workable path, and viewing in a production environment will yield multiple clues about functionality. And crashes.

I stated for our guests, the childish, yet profound Dalai philosophy of PACS: The image has to get into my brain via my eyes, and the interpreation has to get out of my brain and into the microphone via my mouth. That's it! Anything that distracts me from concentrating on the IMAGE is bad, although some interference is necessary, or we're back to film. (Gee, no buttons, no controls, just hold it up to the light...those were the good old days...) So why does Impax 6.0 have, in their words, "enough controls that you could never use all of them if you changed one every day for a year?" (Can you say, "feature fatigue"?) Ah, but this is a deliberate choice! It is easier, I was told, to simply add a multiple-choice option than to tell user A that this item functions this way because user B wanted it that way and not your way. This example was used: if user A wants it to come in black, and user B wants it to come in anything but black, you need to offer it in black and white at the very least. Multiply that by 200+ variables, each with 2-5 settings, and you have a very large number of permutations. Now, the Agfa folks say they found (and were surprised by the discovery) that radiologists have somewhat of the tinkering spirit, and that they all want these variables. My group must be a bunch of primitives, because we all really prefer the simplicity and limited control set of a certain competitor. I think there is a fundamental philosophical difference here. Agfa (and quite frankly Amicas too) is listening to the voices of the buyers on the showroom floor, who are totally enamoured with the 39,000 bells and whistles available on a product which they haven't actually tried in a production setting. Once you take the toy out of the box, you might find that the bells and whistles are so frustrating, it's more fun to play with the box. Feature fatigue, folks, it's a real phenomenon.

Let's deal with some of the issues on my doo-doo list, and see where we are. I am happy to report that a number of those issues were instantly solved by changes in user profile settings. In particular, the problem with the measurement tools disappearing if you click too rapidly turned out to be due to the activation of the 1:1 tool. It assumed that a rapid double-click meant that you want to enlarge a small image to fill the whole screen. Actually, I do like to do that on occasion, but I do NOT like having my ruler disappear in mid-measurement. So, we shut off the 1:1 thing, and my ruler works properly. Why this design was left in place, I don't know.

We actually could not reproduce the situation wherein the wrong prior is selected, so perhaps that was fixed by increasing the number of priors pre-fetched from near-line storage. Whatever. The series bar will be docked at the top of the screen with smaller images, I think. It will work better than it does now in that configuration. Refresh problems may be due to the individual workstation hardware.

The Search problems will be solved in a forthcoming update by making the Simple Search available with Advanced Search afterall. Why wasn't it there before? Well, somebody assumed that clinicians would be the only ones who would want to do simple searches, and that rads would want to do advanced searches, so they made it an either/or proposition. The Agfa folks were rather surprised to learn that the vast majority of my searches are simply by patient name. Just ask me, I'll tell you!

Hanging protocols became an interesting topic on many fronts. I pointed out that the new and improved hanging protocol setup was still impossible to use by the average rad, like me. There was considerable surprise that I would want to do my own hanging protocols. With a bit of frustration, I demonstrated how Amicas does it (I don't think I violated any NDA's....), which is basically drag the series to where you want them, and save this as a named hanging protocol. Wow, what a concept. As it turns out, one of my partners had a similar conversation, and the Agfa guys actually set up a protocol for him. But, once he logged out, and in again, it was gone, something we have seen with some of the other settings. No clue what's wrong here, although there was suspicion on some of the other items that the settings are not circulating between our three parallel production servers. Now, that's another interesting point, if I may digress. I was assured that the whole enterprise could run very nicely on one production server, with one more for testing, and one for fail-over. That's how Amicas does it. No one seems to be quite sure why we went with the belt, suspenders, and staple-gun approach to keeping our pants, I mean our PACS up.

Getting back to a study you have lost is a problem, especially if the machine crashed, because you lose some changes that were made before the crash, or if you use the master "exit" button for that matter. There is a list of the last several studies that you have marked as dictated, which you can access if you happen to know that right-clicking over a particular black spot at the bottom of the viewer will get you there. Rather like playing Doom. Hopefully, there will be some sort of indicator placed here for the unintuitive like me.

The Voxar re-re-re-deployment problem would be helped markedly if the cursor at least showed some indication that Voxar was selected, and I'm told that is an easy fix.

Clone windows and tool toggling are here to stay. The concept of any series in any viewport just didn't seem to impress anyone.

One of the biggest problems we are having is one I neglected to mention in my list, that of multiple people accessing one study. Amicas and even (gasp) Centricity solve this logically: If you touched it first, it's yours, and if someone else tries to get in, they can look, but they can't touch. Not so with Impax. I can be reading happily along with the dicatation status button clicked to "Dictation Started", which should knock the study off of my partners' lists, but it doesn't always, mainly due, I think, to the fact that the worklist hasn't refreshed yet. But instead of telling him to get out of my study, it tells me that he's in it and would I like to give it up? Totally backwards. Add to that, my partner could accidentally click the dicatation button again, and mark the study as dictated, when he just intended to start on it. Try as I might, I could not seem to get the concept across that this was really, really bad for us and needs to be changed ASAP. I finally just said, "I want you to do it like Amicas does it. Period." It seems, that some of Agfa's academic customers want the ability for a resident to start a dictation, and an attending to finish it, so they don't want to lock out this function. Well, I was just told that the easiest way to keep your customers happy is to give them all the options they want. Fine. I absolutely want to be able to lock down a study in the manner described above. Please make that happen. And by all means, give the academs what they want, too, but let me turn it off.

And so, I'm back to where I started. I really need to say that Impax 6.x is a monumental achievement, despite it's flaws. To port a huge program from C++ over to .NET and to make it web-deployable ain't easy. (I say that like I've done it 17 times last week....) The Agfa team should be very proud of what it has accomplished. Now, let's build on what we started this week. Impax 6 was developed somewhat in a vacuum, that is, there was not adequate influence from users of other products, and there was not adequate testing in a production environment. What we did this week should be absolutely mandatory for all new PACS products: have the programmers and such sit next to a radiologist while he uses the system, and get instant feedback for improvement. Do this with at least 100 rads, and not only will you iron out the bugs, but you will discover newer and better ways to do things. And you may find out that toggling of buttons, and text screens, and cloning of windows just isn't the best solution.

So, what do I say now to potential buyers? Frankly, I think you need to wait, although I am very confident that our crashes will be fixed, and that the repairs and upgrades outlined above will happen in a very timely manner. I will let you know as soon as this occurs. In the meantime, Happy Thanksgiving, eh?

1 comment:

Anonymous said...

It sounds like you need a good system administrator to sit with you for alot of your issues.