Impact of PACS Federation on Hospital Network Traffic
 
Authors:
Rasu B. Shrestha, MD, MBA, University of Pittsburg Medical Center; Nathan Lauffer, MS; Brian Kolowitz, MBA; Dustin Schultz; Gonzalo R. Lauro, MBA; Harry Black, MBA
 
Background:
The University of Pittsburgh Medical Center is spread across a wide geographical area and comprises 20 hospitals and 30 imaging centers. Due to historical reasons and performance concerns, UPMC’s PACS landscape consists of 12 separate installations of PACS, which until now, were distinct silos of information. We have created a standards-based platform, called SingleView, which federates the PACS enterprise and provides, for the first time, a unified view of all of a patient’s priors and reports across all hospitals in the federation.
 
Evaluation:
UPMC’s 20 hospitals are connected at various network bandwidths, ranging from 1GB Fiber, to 50 MB SES, and Point to Point T3. With the creation of a PACS federation, radiologists and clinicians were more aware of prior studies and reports, resulting in increased access to those datasets across all hospitals. Several models were created to predict increased network traffic and to evaluate the impact on the Enterprise WAN, and actual measurements of increased network traffic were performed after the implementation of the federated model. We also evaluated typical usage of imaging, with and without SingleView, over several 24 hour periods across several hospitals (Figure 1). Quality of Service (QoS) service levels were applied across critical data, bulk data, network control, and video data. Tools such as NetScout Performance Manager and OpNet were used to closely track the impact of the federated PACS (Figure 2).

Figure 1

Figure 2
 
Discussion:
On average, 23.1% of patients’ studies were found to have relevant priors in vaults, other than the primary vault (Figure 3). Prior to SingleView, radiologists and clinicians were literally unaware of the existence or the location of those prior images and reports. Consequently, patients would get additional, unnecessary tests performed, or would get a suboptimal report on a current study. In this paper, we discuss the network impact across the range of bandwidth environments in the core and community hospitals, the lessons learned in deploying a federated PACS model across a range of connections, and share nuggets of information on architecting a broader deployment of imaging services across hospitals.

Figure 3
 
Conclusion:
SingleView has provided a seamless platform of workflow integration across multiple PACS implementations in an era of distributed diagnostic imaging even across a varied WAN landscape.