Streaming technology is the future made present to those who want to watch movies anywhere at any time. It’s becoming the same thing to radiologists, clinicians and patients who want mobile access to medical images.
“How many people bother with renting or purchasing DVDs in an age of Amazon Fire TV or Roku?” wrote Joseph Marion, a healthcare technology consultant whose firm is Wisconsin-based Healthcare Integration Strategies. “The same may likely be the case for digital image access and interpretation through streaming technology.”
For provider organizations that implement vendor-neutral archives, he continued, adding a server-side rendering (SSR) solution— which allows a user’s browser to fetch data from a program or application running on a server rather than one running on the user’s computer—“can augment the need to acquire a complete PACS solution.”
Intrigued by that assessment, this reporter reached out to Marion with questions on the background, basics and finer points of SSR technology.
What technical and economic developments facilitated the rise of SSR in medical imaging?
Joseph Marion: From a clinical perspective, the key driver probably was in the advanced visualization area. People were looking for more cost-effective solutions. If all you needed was a single advanced-visualization workstation, then you probably didn’t need server-side rendering. But server-side rendering can offer some significant advantages to some of the larger provider organizations and integrated delivery networks.
From a technology point of view, healthcare has always capitalized on commercial or consumer developments. And as graphics processing units became more powerful and sophisticated, those advancements became more applicable to applications in medical imaging.
In terms of what’s attractive about SSR, it’s the speed and the potential to lower costs. And from a data security point of view, the fact that no information is being transmitted to the clients is another major advantage.
The only remaining question is on latency of operation (the time it takes to send a message from the client to the server and then back again). So bandwidth is going to be a significant factor in the context of whether it’s an appropriate technology or not.
In what scenarios is SSR clearly superior to client-side, where it’s the local computer or workstation that runs the application? In what scenarios is SSR marginally superior or not the best choice at all?
When you have multiple users in varied locations, there is a clear advantage to SSR technology, because it does offer “any image, anywhere, anytime” kind of access. This is appealing to some of these larger multi-site integrated delivery networks.
If all I have to do is keep a server up to date and any web browser can access it, then I am simplifying how I maintain and update the capability as well. Which again is appealing to some of these larger provider organizations.
What constitutes good performance in server-side rendering?
The gold standard basis was any image, anywhere at any time—and within two seconds. I think that’s probably the benchmark. If you can put the images up there in two seconds for a diagnostician, that’s acceptable performance from a clinical perspective.
And then for clinicians, clearly, they maybe don’t need quite the same level of performance as the diagnostician does. And the patient definitely doesn’t need the same level of performance. So you can accept or tolerate more latency in those situations then you could for the diagnosticians.
Are there any particular technical challenges IT departments should anticipate when moving to SSR?
I think one of the clear questions is going to be, if I am going to locate this thing centrally, do I have any redundancy in the system? Because then I am dependent on two factors. One being if I only have one data center where the servers are located, I have no redundancy. If it goes down everybody is down. And then secondarily, dependent on the network in the context that if the network goes down, likewise, I have no accessibility.
So I think those are probably factors that facilities would need to look at. Do they have the infrastructure in place to be able to support redundancy?
We hear a lot these days about zero footprint viewers. To what extent are these a part of SSR and to what extent are they a category of their own?
There are two pieces to a server-side rendering implementation. There is the rendering engine, if you will; the