AuntMinnie.com · The PACSMan Pontificates, Episode 4: The perfect PACS
I had a client who after two less-than-perfect PACS implementations in a decade said that anything less than a perfect PACS next time is unacceptable. Of course this same person is on his third marriage as well so perfection may well be in the eye of the beholder…LOL.
Is there a perfect PACS? Sorta, kinda, but not really. Just like there is a perfect marriage, it may exist but most are far from what many would call perfect or even near perfect. Many would say the 63-year long marriage of Ruth Bell Graham to renowned evangelist Billy Graham defined ideal if there ever was one.
When asked if she had ever contemplated divorce during their 63-year marriage Mrs. Graham famously fired back, “Not once. Murder, a few times—but not divorce!” I run into so many sites that, at the very least, contemplate divorce almost daily and would escalate it to murder if they found the right person or persons within the company to deal with the challenges they faced and thought they could get away with it.
When the honeymoon ends
Like all marriages, newly implemented PACS are great initially -- much better than what you had before. Once the honeymoon is over though you tend to see things that you conveniently overlooked during the dating phase. It’s still better than what you had but not quite as perfect as you initially thought. You knew what you liked initially –- it was all you wanted and needed -- but didn’t realize that the carry-on bag that came with the system initially was followed by 3 extra pieces of checked luggage once you spent more time with the system.
It all starts before a contract is even signed. How receptive a company is answering questions beyond what is shown on the web site that are both application and product specific is huge. This is especially important if the answers are ones provided in non-technical language.
Once a contract has been signed establishing an implementation timeframe is critical. Typically a vendor needs 30 days or so to evaluate internal and external resources and to coordinate these with the client. Unlike hardware-based systems, using the cloud eliminates the time it takes for hardware to ship in, be configured, tested then shipped to the customer site.
Still a limited amount of hardware is required although the amount is fairly limited. A lot of what is needed is COTS (commercial off the shelf) hardware and sourced fairly easily. This is especially important if parts of the existing radiologist workstations will be reused or additional ones are required.
Because an implementation takes less time it does not mean it’s better. Implementations rarely meet the proposed schedule with most taking at least three to six months past the promised date to be ready to be signed off on.
Failure to communicate
Most of this relates to the failing of both parties to fully disclosing their expectations and the vendor understanding them as well. I have banned the words “I thought…” from both clients and vendors I worked with. Just like in the movie Cool Hand Luke, the Captain may tell Luke there was a “failure to communicate,” but like in the movie that failure in real life comes with consequences as well.
You would think contracts, at least in theory, should address most, if not all, of the contingencies that pop up but in reality most of the time they don’t. Oftentimes the solutions aren’t quick or easy as well either. The PACS may be perfect from what you saw initially -- a first, third, or even fifth date that still has you enamored with them -- but just like dating the longer you are with them the more deficiencies you tend to find.
Some you can live with and overlook while others just grind on you. Newer PACS that at least on the surface seem more perfect become an option especially when “cheaper to keep her” is no longer something viable. Faster better cheaper becomes the new mantra..
About a third of the PACS people who get PACS now are replacement PACS because hardware or software is no longer supported and needs to be replaced. Another third are selected because the existing system has a negative impact on workflow and overall radiology operations. This is often due to increases in volume.
The latest and greatest
The final third are those wishing the latest and greatest solution from using the Cloud to an integrated AI solution. Technically there is nothing wrong with the existing system besides being a bit slow but they want it better or to make it better because they read somewhere in some journal about how XYZ has a crazy impact on productivity and now they want that.
The reality is the only studies have you can take to the bank are those which impact the reporting process. The rest show benefits for sure but not a strong return on investment (ROI) like so many claim. This is especially true of AI.
Turning over much of the PACS support to the vendor when it is hosted in the cloud has a bunch of benefits. Not the least of these is not having to deal with hardware support any more or replacing it when the software a new operating system is required like Windows 11 recently was.
Updates and upgrades now come at regular intervals and you get them whether you want them or not, Most don’t require prior testing on the end user side, uptime increases (excluding Cloud service provider (CSP) downtime), distributed data throughout the enterprise is a given, easier operation that increases productivity is a plus and others round out the list. But perfection has a price and cost. The local area network (LAN) may need to be upgraded to support the speed desired and expected for high slice count studies now being done.
A significant advantage
The CSP has a significant advantage associated with it as does cloud archiving. Having the archive in the cloud, as long as images are stored in a DICOM file format, significantly speeds up the retrieval time from having to go to an on-site vendor neutral archive (VNA) and allows for backups to also be created at a much lower cost.
Training can take several weeks if not months especially for applications that are not used often. In those situations productivity can actually decrease. There are always tradeoffs. Sadly what often happens is the application you loved and that provided you with one of the justifications for the purchase gets deleted with the next upgrade or update. You loved it before and now….. It’s gone with no reason given. It’s like doing the wild thing several times a week when you were dating and then wondering why things have slowed down now that you are married.
Know that if not many use an application it can be gone in a New York minute. Change is the only constant. Yes, you are still in love but….
One of the biggest frustrations with PACS is when it doesn’t work or work right. With an on-prem solution you can call into the support team and get help. It may take an hour or a day to fix it but either one equates to an eternity if you are a radiologist trying to get through a worklist. That is one reason why the perfect PACS is a hybrid solution, with the primary PACS residing in the Cloud with its own backup and secondary/backup either an on-prem solution or one residing in the Cloud.
A hybrid system
Backups are great but if images do not get to the Cloud to be distributed the Cloud is basically worthless unless you have a way to get the images direct from the facility. This requires a hybrid system -- Cloud and on-prem combined.
When the unexpected happens, Cloud providers often have strategies in place to handle those situations. They don't just sit around biting their nails, hoping for the best. Most have built systems that help them manage outages and downtime wisely. For starters, many of these providers use something called redundancy. This means they have extra resources available, just waiting in the wings to step in if something goes wrong.
Sort of like an understudy at a play. If one server goes offline, there’s another ready to take its place instantly. This minimizes any downtime you might experience because the switch happens so swiftly that you often won’t even notice it.
Some cloud providers work with global data centers. Imagine having copies of your data stored in several locations around the world. If there's a blip in one area due to a natural disaster, a power failure, or even a tech-related issue, your data is still safe and accessible from another spot. It’s like having a safety net that’s designed to catch you in case you fall.
And because they have all these data centers, they can quickly redirect traffic to ensure that everything keeps running smoothly. It’s a clever way to handle issues that come up without you needing to worry about losing access to anything important as long as adequate security is provided.
Once again though, you need to be able to get images from the site to the Cloud to make it work. If you have redundant connectivity to the Cloud, if one line gets cut you can always switch to another. But like everything that comes with a price.
Load balancing
Load balancing is another trick providers use. When traffic hits a particular service hard, load balancing ensures that the workload is spread across different servers. When there’s an unexpected surge in use, instead of one server taking all the heat, the load is shared. It makes it a lot harder for any one server to get overwhelmed and crash.
I mean, it’s like running a restaurant with a crowd: if one waiter is overwhelmed, you don’t just let them drown while the rest of the staff stands by; you send help! That’s how providers try to keep everything running without a hitch.
Outages can happen at any time, but the best providers are the ones that take a proactive approach and don’t wait for a problem to rear its ugly head before they act. They constantly monitor their systems. They may have teams of people tracking performance data in real-time, checking for anomalies, and predicting issues before they escalate.
AI can play a role here as well by learning to identify trends and problems faster. If a potential failure is detected, their systems can often react automatically to fix the issue without us even knowing one existed...
Downtime
Downtime isn’t a complete black-and-white scenario either. It’s not like everything either works perfectly, or it’s totally offline. Providers categorize downtime, and they have different strategies tailored to different types of incidents. If it’s a minor issue, they may fix it on the back-end without affecting users. But if it’s a large-scale problem, they typically have a full communication strategy in place.
Communication is essential during an outage. A lot of the top-tier providers send out updates to their customers, letting them know what’s happening, what they’re doing about it and any timelines for fixing the issue. It’s like dealing with a power outage in your neighborhood; you don’t want to be sitting in the dark wondering what’s going on. They have status pages that you can check or send messages to your cell phone, which is a great way to keep everyone in the loop.
But let's be honest, outages can be pretty frustrating, especially with my local cable provider. A rad might be in the middle of an extensive study and then boom lose the connection. You can’t eliminate the risk entirely so you focus on recovery after an outage. That includes things like back-up systems and plans for restoring services as quickly as possible. Some vendors put a back-up on-prem solution that replicates what is being sent to the Cloud server so if the connection to the server is lost they can still read studies being done.
Depending on how it is configured you can even store a small amount of priors on this unit as well. Rads would just dial in and read studies from here until the system on the CSP is restored. Those are the best designed systems so that can be minutes hours or days don’t really matter depending on the severity of the issue. Having a system designed like this also eliminates the cost of a separate high-speed line to the CSP as well.
Compliance and regulation
Compliance and regulation is another layer to consider. Many cloud providers have to adhere to specific standards and practices, especially when dealing with sensitive data without compromising data security. Double-checking their protocols and ensuring they’re back on the right track once everything gets restored is crucial.
Cloud service operation has a lot to do with the end user as well. Most CSPs have a service level agreement (SLA) that outlines their commitment to uptime and what compensation you might be entitled to in cases of prolonged outages. It’s basically a contract that sets the expectations for both sides. Sadly, like most PACS contracts collecting on the SLA isn’t easy to do.
Cloud computing is a wild ride, and while outages are part of the journey it’s vital to see how providers handle them. There’s a lot going on behind the scenes to manage those hiccups. It's also a great reminder that technology, while fantastic, is imperfect, and we can all benefit from being prepared.
So do your homework and do your best. You probably won’t see perfection in the lifetime but most people would be happy just with pretty darn good.
Michael J. Cannavo is known industry-wide as the PACSMan. After several decades as an independent PACS consultant, he worked as both a strategic accounts manager and solutions architect with two major PACS vendors. He has now made it back safely from the dark side and is sharing his observations.
His healthcare consulting services for end users include PACS optimization services, system upgrade and proposal reviews, contract reviews, and other areas. The PACSMan is also working with imaging and IT vendors to develop market-focused messaging as well as sales training programs. He can be reached at [email protected] or by phone at 407-359-0191.
The comments and observations expressed are those of the author and do not necessarily reflect the opinions of AuntMinnie.com.
