I obviously have a knack for getting on paper what a lot of people have thought and didn't realize they thought. And they say, 'Hey, yeah!' And they like that.
I don't pick subjects as much as they pick me.
I'm not sure how much of a knack I have for anything
these days, but I've always been inspired by the gentle, yet biting humor of the late Andy Rooney. And so, this piece is dedicated to his memory.
"Did you ever wonder why...?" The opening clause that sunk a thousand fools, righted a thousand wrongs, and warmed the hearts of thousands. Today, I'm asking you, my dedicated readers, both of you..."Did you ever wonder why we can't make a better PACS?" And I'm hearing you respond, "Hey, yeah! Good question, Dalai!"
Let's try to answer it.
Vegetarian - that's an old Indian word meaning lousy hunter.
I don't want to toss out a blanket indictment of every PACS from every vendor. Clearly, most
of them work most
of the time and do most
of what we want them to do. So we settle and tell ourselves we didn't really
want our PACS to function any better than it does. But I think we are still left with the vague feeling that things could be improved. A lot. And in fact, I'm not alone in this impression. Bruce J. Hillman, MD, and Bhavik J. Pandya, PharmD are in full agreement after conduction a survey on the topic. As reported in JACR
, they found:
All 5 respondents pointed to the lack of intuitiveness of their systems as causing them inefficiencies and fatigue. Their concerns in this regard centered largely on their current workstations not easily presenting them with the full range of tools and options available, and variability among the different user interfaces of the workstations they used daily.
Viewing our results qualitatively, there is a convergence of opinion among the respondents about the key shortcomings of their current PACS. The substance of what each radiologist told us was the same: flaws in design and IT connectivity diminish radiologists' productivity.
I could have told you that without conducting a study. Well, I guess you could say I have
conducted a personal study over the past 20 years of using PACS in one form or another, and I certainly agree with their observations. As Andy Rooney put it:
People will generally accept facts as truth only if the facts agree with what they already believe.
My good friend Brad Levin has one of the finest resumes in the PACS business. I'm proud to say I've known him since the AMICAS days, and I'm grateful that he is still
is willing to talk to me. Brad published this list of PACS grievances in response to the proposition that "Radiology has solved the problems of going digital." Ha! Anyone who believes that
hasn't touched a PACS interface. Rooney's take would have been:
The 50-50-90 rule: Anytime you have a 50-50 chance of getting something right, there's a 90% probability you'll get it wrong.
In many ways, PACS makers got a lot of things right. But they got a lot wrong and continue to do so. Brad's list
, which I have abridged below, was gleaned from a survey he conducted. It reads a lot like my Laws of PACS:
Hanging protocols. Nearly every PACS has them, but how many work as expected? A huge number of systems I’ve encountered have flat‐out given up...
Timely access to priors. Many organizations are still routing the same DICOM studies to multiple destinations because they don’t know ‘who’ is going to interpret the studies. It takes so long to move the DICOM, they can’t afford not to have the images at the right location...I’ve also run into systems that have multiple hospitals using the same vendor PACS, and yet they still do not have access to priors due to a variety of technical barriers.
Viewer overload. A multiplicity of viewers exist at most systems, especially those that have grown over the years. Viewers for the radiologists, for the clinicians, for access from the EMR, for access from CD/DVDs, for QA stations, etc. That’s 5 viewers...
Viewer capability. Radiologists and referring physicians are using viewers that were designed years ago to take on today’s challenges...
Won’t work. Does this sound familiar? “A prominent referring physician’s office just upgraded all of their PCs. They finally got rid of their old clunker systems, and they are now running the latest Windows 8 systems. They are literally on cloud nine, and we just told them our image viewer does not support the latest release from Microsoft.” You can also substitute, “We just upgraded our offices with new Apple iMacs.” This is bad customer service 101.
The reality is the vast majority of PACS in use today are woefully lagging behind their support of the latest operating systems, web browsers and platform support (e.g., Mac). It’s no wonder many referring physicians are frustrated with Imaging.
Advanced visualization. In 2011, KLAS reported that Radiology had not found an effective way to work 3D imaging into the workflow of the radiology department. I see this everywhere ...
...(I)t is common to see studies such as PET/CT and CTA only available at isolated workstations. If the radiologist is not at that specific station, they do not have access to the images. Far too frequently radiologists are forced to move to the images. That’s an archaic practice in today’s high‐tech, mobile world.
Speed of access. As mentioned, the majority of today’s diagnostic workstations and clinical viewers were originally designed a decade ago or more. When those old viewers were forced to support multi‐slice CT in the mid‐2000s, it took several years for viewer performance improvements to catch up. But the growth in multi‐slice studies has continued in terms of study size and number of slices. One prestigious system out West has a current benchmark that their viewer(s) need to be able to support rapid local and remote access to current + (multiple) prior studies totaling 8,000 slices. If viewers don’t support 2‐3 second access, they are no longer being considered...
Remote/At home access. The PACS revolution eliminated film, but an embarrassingly large number of institutions to this day do not provide radiologists the same level of access at home as they provide at the hospital or imaging center that they work at during the day. The legacy technology either is too expensive to support from home, or does not provide adequate speed/quality of access over consumer networks using VPN. As many institutions strive to take‐back‐the‐night, this problem needs to be solved.
Mobile access and image exchange. Despite the availability of mobility and image exchange solutions over the past several years, the use of these solutions is far too low in actual practice. My guess is hundreds of facilities are using mobility and image exchange solutions, when they should be in use at thousands of facilities.
Unsustainable workflow. I’ve seen each of these reading workflows at multiple settings, from coast to coast ‐‐‐ Swivel‐chair workflow: A radiology group reads for multiple entities, each with their own RIS and PACS. Today’s typically used solution is to have a dedicated workstation for each entity and literally have the radiologist move in the swivel‐chair, from one station to the next, to read the day’s studies...Literally a setup of workstation overload, to perform multi‐modality analyses, instead of reading off a single viewer.
Graphically, survey SAYS (in the tone of the late Richard Dawson):
|Image courtesy siimcenter.org|
If I were in the PACS business, I would hide my head in shame.
In the end, we are dealing with two
intertwined problems, the PACS architecture and the graphic user interface, or GUI. Both are languishing somewhere in the late 20th century, and thus, so are we.
With respect to the GUI, Dr. Elliot Fishman
, whom some have called the World's Best Radiologist, lays it on the line:
As I sit here at my PACS workstation, I see a long list of icons on the left, most of which I have neither ever used nor know what they do. Our newest 3D imaging system boasts a bevy of icons that are little more than symbols—possibly only recognizable by cavemen—and unexplainable motions for the right and left mouse button. It makes one wonder why things aren’t simpler, similar to what we see on an Apple iMac or MacBook, or iPad, or the user interface screen of Amazon.com.
The aforementioned are examples of technology used by millions and “customized for every one of them.” It seems that lessons learned there have never made it into the medical arena, let alone radiology. Why must my PACS screen look exactly like yours, especially when we recognize the inefficiency that comes with the lack of customization? How is it that Amazon remembers every purchase I ever made and makes suggestions for what I might want or need, while my PACS workstation acts every morning as if we’ve never met before? Why is it that evolutionary and revolutionary changes in Google and Facebook continually affect everyone and yet those changes never make it through to how we practice radiology?
More on this in a moment.
We need people who can actually do things. We have too many bosses and too few workers.
As an aside, some have suggested
that IT-savvy departments assemble their own PACS from off-the-shelf components. To that, I can only say, "BWWWWAAAAAHAHAHAHAHA!" Good luck, folks. Not going to happen for the foreseeable future, at least not in my enterprise.
The world must be filled with unsuccessful musical careers like mine, and it's probably a good thing. We don't need a lot of bad musicians filling the air with unnecessary sounds. Some of the professionals are bad enough.
I'm not going to get too deep into the architecture discussion, as many of you could talk rings around me. There are a number of ways to skin this particular cat, and technology will provide the answer. What we know is that the old, distributed architecture from the 1980's no longer is satisfactory, and hasn't been for quite a while. The web-server model, more or less the de facto standard today, can't cut it anymore, as demonstrated by the various problems Brad outlined above. To some degree, the problem is bandwidth. 8000 images might be roughly 4 Gb of data, and if you're sitting in a Gigabit Ethernet environment, we're talking 4 second delivery. Lossless compression brings this well within Brad's tolerances. But this doesn't do much for the home or mobile environments. I just upgraded my U-verse home internet to 50 Mb/sec, and on AT&T's LTE, my iPhone can reach 80 Mb/second. Streaming and compression will help, since we aren't going to see gigabit speeds outside the institution until we get the next wave of bandwidth innovation (5G? 6G? Fiber to the home?).
The answer here is probably server-side rendering; don't mess with sending the data at all. But this is such a huge paradigm change, you don't see many PACS vendors doing it. Basically, the number-crunching gets done on Big Computer in the data center, cloud, Mars, wherever, and we only see the pitchurs. (Of course, the remote site has to have at least barely adequate bandwidth, 4Mbs or so, on both the uplink and downlink sides. Our IMPAX requires each command and mouse stroke to be transmitted back to the production server, and this has led to slowdowns even with
This goes hand-in-hand with the so-called zero-footprint viewer. You use your computer and browser to peer into
the system, and none of the data ever is truly on
your computer. This certainly helps with security concerns, and solves multiple problems, not the least of which is access on devices running something other than a particular older version of Windows that your particular PACS software demands for its particular fetish. To show you how far we haven't
come, I remember the days when you could not access most PACS via the internet. Mitra, now part of Agfa, came up with one of the first ways to do so, an appended web-server that had to be grafted onto the PACS called the Web1000. Today, to get iPad access for our Merge PACS, we would have to get an appended zero-footprint viewer server called iConnect, which is too expensive to justify for that purpose alone.
Making duplicate copies and computer printouts of things no one wanted even one of in the first place is giving America a new sense of purpose.
The answer to Elliot's question of why things are as they are is rather simple. We, the radiologists and technologists, the actual end-users of these products are in general, NOT the decision-makers on their purchase. More often than not, the IT department, that has little to no understanding of what we do and how we do it, chooses the PACS vendor based on how easy said vendor will make their
lives, but not ours
. This corporate mentality has to change. Elliot concluded:
We need to find a way to encourage those companies that are designing the future—like Apple, Google, Amazon and Facebook—to help us create our future. I think it is neces- sary not only for our survival but also if we want to continue to be innovators in patient care.
Or, at the very least, we need to figure out what those companies are doing right, and get the PACS vendors to implement it.
I'm a big Apple fan. Between me, Mrs. Dalai, Dolly, and Dalai, Jr., we have one iMac, two Macbook Pro's with Retina screen, one regular old Macbook Pro, three iPads, and four iPhones. Not counting the two old Macbooks and a dead iPod or two sitting in the closet.
I don't like food that's too carefully arranged; it makes me think that the chef is spending too much time arranging and not enough time cooking. If I wanted a picture I'd buy a painting.
What is the secret to Apple's iSuccess? Some have called this "The Humane Interface"
A key to Apple’s success is the company’s insistence on reducing options in the name of reducing complexity. Those who decry Apple customers as fanboys attack us and the company alike, saying that because Apple chooses to focus on simplicity, we and it must also be simple. That’s the wrong interpretation of the facts. Instead, Apple’s focus on simplicity isn’t about reducing choices to make computing idiot-proof; it’s about focusing on the important bits instead.
It wasn’t the first iMac that came along and disrupted things. It wasn’t even Mac OS X. It was the iPod, and even then, not all at once.
The iPod’s true advantage was that it was just easier to use. It had fewer buttons, looked nicer, synced with iTunes, and was the only music player at the time that could play songs from the iTunes Music Store...(Y)ou had to beat the entire experience, not just the device.
Design is a series of decisions. Should it be this color or that color? What’s the first thing you see when you log in? What happens when the user clicks here?
Sometimes these questions are really hard to answer, and the easy solution is to make it a preference for the user to decide instead. But the best designers tend to view such options as admissions of failure. Where Apple differs from its competition isn’t in aesthetic beauty, it’s in the company’s ability and willingness to make decisions on behalf of its users.
(T)he megahertz race is over, and it was won by the people who just wanted to check their email and surf the Web without having to think too hard about what they were doing.
While RIM was busy making BlackBerries that appealed to network administrators, the people who actually have to use the things were going out and buying iPhones. No surprise, then, that the next great leap forward in technology was the removal of the keyboard and mouse. What could be more human than touch?
Linux and its cousin Android win with hobbyists and technology enthusiasts by providing options for everything. Like software development itself, the use of an application becomes a flow-chart of possibilities. Where, then, is the line between configuration and programming?
Apple’s take is to remove complexity and make choices long before the user sees the product. For some, this feels like control is being taken away, and they accuse Apple of dumbing down their products, presumably giving us the old cliché that Apple products are for dumb people. For those of us who prefer technology with a human touch, the magic is in what we can accomplish. Our tools are extensions—not reflections—of ourselves.
It turns out that the real secret to making computers usable is to make them disappear. Our humanity is finally catching up with our technology.
In these paragraphs is the key to the future of PACS. Very simply, I was right years ago when I drafted the Laws of PACS
. I'm thinking specifically of the Fourth Law: "PACS should not get in your way." I can't say with certainty what a Zen/Steve Jobs inspired PACS GUI might look like, beyond simple, well, simplicity. PACS should anticipate the tools we need and provide them, hiding the other 100 tools and buttons that we don't
need at the time. Displays should be fluid and adapt to the task at hand. Stuff like patient-demographics and lab results need to be unobtrusively available; basically, we need a transparent window into the EMR. 3D displays need to come up as part of a super hanging-protocol if you will. In essence, the darn thing needs to be intuitive, as we've said above. And trainable too, adapting on a case-by case basis to your needs.
Such easy concepts, so difficult to execute.
We're all proud of making little mistakes. It gives us the feeling we don't make any big ones.
So how do we get from here to there? One big impediment to progress was outlined above: those who buy
the PACS aren't those who use
the PACS. So should we go on strike until someone listens?
Let's make a statement to the airlines just to get their attention. We'll pick a week next year and we'll all agree not to go anywhere for seven days.
I really don't have a better idea. And besides:
I'm in a position of feeling secure enough so that I can say what I think is right and if so many people think it's wrong that I get fired, well, I've got enough to eat.
Thanks, Andy. We miss you.