Archiving and accessing medical images is an increasingly complex issue, and one that touches a rapidly growing body of caregivers—which is why cloud-based archiving, in spite of lingering disadvantages, will be the model of the future. That is the prediction of James Philbin, PhD, senior director of medical imaging at Johns Hopkins University, Baltimore, Maryland. Philbin presented his perspective on the future of cloud-based archiving in a June 8 session at the Society for Imaging Informatics in Medicine (SIIM) 2012 meeting entitled “Next Generation Archives: Local vs. Enterprise vs. Cloud; VNA vs. ANV.”
Philbin notes that “cloud” has become something of an all-purpose term. “Whatever you imagine the cloud is, it seems to be,” he says. For medical image archiving and sharing, however, he narrows down the definition to mean CPU and storage virtualization services provided by a third party from a remote infrastructure. Functionally, he observes, there should be little distinguishing a cloud-based archive from one hosted in an on-site datacenter. “The key differentiator for the long term is who is managing the system,” he notes. With that differentiator, however, come critical questions about data control and security.
Weighing the Benefits
The promises of the cloud are myriad, Philbin says; among them are a more economically viable, pay-for-what-you-use cost structure and the scalability to add or subtract resources on the fly. Other advantages include high availability and redundancy; anything stored in the cloud is likely backed up to multiple datacenters, offering advantages from the standpoints of disaster recovery and business continuity. He notes that the financial industry already effectively leverages this form of archiving for sensitive data, but adds, “I think it’s going to take us a while to get there.”
Why? For one thing, cloud-based archiving presents problems of its own. User-side hardware “edges” to the cloud are necessary to preserve data in the event of a network outage, and users also have concerns about performance when it comes to uploading and visualizing studies, particularly in rural areas where Internet connections are not as robust.
The biggest problem, however, is more ephemeral, Philbin says; it centers around CIOs’ loss of physical control over where their data are warehoused. “It’s a real issue,” he says. “We’re really going to have to trust and have strong legal agreements with cloud providers.” As it is almost impossible for a CIO to know what’s happening in a remote datacenter managed by a third party, Philbin calls for good controls, reporting mechanisms, and dashboards for cloud-based archives to assuage in-house IT staff concerns.
Security of patient health information is also a critical concern, but here, Philbin feels that cloud-based archiving has an edge. As he points out, cloud providers have a strong incentive to make their datacenters as secure as possible, and can leverage their scale to invest more heavily in security measures. “The business drivers will compel them to be better at security than we could ever hope to be,” he predicts.
Philbin notes that several emerging technologies will help push health care providers toward cloud-based archiving by enhancing its viability as an alternative to traditional datacenters. Zero-client applications are proliferating in the health care IT sphere, allowing, for instance, Web-based diagnostic viewers and workstation-free advanced visualization. These technologies once required dedicated software, but now, with the right Internet connectivity, could be accessed as services provided through the cloud, especially as GPU virtualization becomes more commonplace in the years to come.
Increasingly, Philbin notes, health care providers also will use virtual desktops for diagnostic workstations, improving their security while lowering total costs. “This is a generic way to remotely render a legacy thick client,” he observes. Both the virtualized and the zero-client approaches can be used locally, but they also can be used in the cloud.
As Philbin notes, today radiologists are often forced to diagnose using incomplete information, because prior studies are often spread over multiple health care providers and impossible to access. The results are redundant studies, extra radiation dose, and increased costs—all factors that