SOA Breaks Up Radiology’s Toolkit

In healthcare there is a lot of interoperability missing. This is rooted in how data capture has been done and how data collection from legacy systems has been integrated. The true value for using service-oriented architecture (SOA) is in the sense that you can build out live and standards-based Web services that can make the integration process happen.
How to decompose a silo-architecture PACS – Taking radiology as an example, PACS has basically two core functions - image storage and image discovery. The storage service does not have to be in the hospital. It can easily be out-sourced; the ASP model for offsite storage follows this model (image repository). This is essentially already a kind of storage service in the concept of SOA. Discovery is a push or pull-based image query and retrieval workflow of images. Discovery requires an index service (a registry similar to IHE-XDS/I) and a set of business rules.
Decoupling the two PACS service components allows workflows between services, access to multiple distributed services at a time and a reusable services framework, typically independent from the hosting environment (e.g., OS, hardware, etc.). One question that arises is how to interface various services and applications (aggregation of services) in a standardized way. The Web services concept that uses the Simple Object Access Protocol (SOAP) presents a standard interface layer. This interface layer can be used in radiology for storing, query and retrieve operations. In addition one can easily imagine an SOA-based image processing service, reusing image publication and discovery services from the PACS services. The processing might be internally proprietary, but because it uses a standard Web service interface, it can be discovered and used by a larger community - in which case other institutions can share that resource from the Internet and optimize utilization.
Services can be provided as enterprise services within the hospital network or as wide area network services (WAN) outside on the Internet. As such, SOA completely breaks up the traditional silo-architecture structure within the radiology department, which is typically vendor and modality oriented.
Operations - From an operational point of view SOA makes good sense because it allows dynamic resource discovery and allocation (e.g., grid technologies advocate this feature), advocates use of standards based interfaces, enables resource publication and thus sharing for increased economic use, and enables an open marketplace for innovative services (e.g. fault-tolerance, disaster recovery).
Networks Enable SOA - It just depends on network connectivity. In the last couple of years, network resources have become more available. That means the opportunity for hospitals and medical centers to actually utilize high-bandwidth/lower latency networks, which opens the possibility we need for network-based SOA going into healthcare. That enables SOA resources to exist on the Internet and be available as individual service components. You can think of a PACS as being a combination of a discovery storage service and a workflow engine, which is another service. You have to think about RIS as well. While not all features found in RIS today might be useful to expose as services, such as local scheduling, others like transcription services would be.
Is this is a threat to vendors? – Not necessarily, if vendors rethink their model in time. The PACS industry depends on multiple technology upgrades, so this is just another one.
Decoupling of Services – There will be more separation between the actual storage provider and PACS software provider. There is already a movement away from the silo-architecture system where one vendor provides all of the components within a PACS. A couple of storage providers, like IBM and HP, have partnered with other PACS vendors to provide storage functionality, while PACS vendor only provide the software running on the third-party hardware. I think that these vendors are more likely to be successful in a SOA model because they are already breaking up specific components into internal services.
Radiology Enters a New Paradigm - For radiology, this changes the way in-house resources have to be looked at and it creates a huge dependency, as a trade off, on the network environment. Most of the hospitals use T1-type Internet connections and rarely use this as part of their operational component. Leaning toward SOA will change emphasis on network resources. SOA provides an opportunity for hospitals to think about what list and aggregation of services they need to be operational, in other words, in radiology operations. We already see decoupling of what is an internal and external service. Many hospitals already use outside radiologists, which is in some sense another service in the SOA model – looking for outside services and decoupling function into a set of individual services.