Deployment of a Diagnostic Imaging Repository (DIR)
for the Regional Storage and Sharing of Medical Imaging Data
in a Heterogeous RIS-PACS Environment
 
Authors:
Alain Gauvin, MSc, Bureau du Dossier de santé du Québec; Greg Ruthman, BSc; Suzanne Laframboise; Richard Blouin, BSc
 
Background:
The regional consolidation of medical imaging data is contemplated in many jurisdictions world-wide, but has been carried out in fewer of them. The reasons for accomplishing such integrations are generally centered around two main needs: sharing those images from a central location, and consolidating the long term storage for those data to realize economies of scale in the initial and recurrent costs associated with the storage of the large volumes of data typical in medical imaging. The deployment of an edge-server-based DIR to support interfacing with a vastly diverse group of RIS/PACS solutions is described. The DIR in question has been installed to receive and store the imaging data produced in a multi-regional setting with a yearly production comprised between 6 and 7 million studies.

At the beginning of the project, 6 different PACS vendors, along with 5 RIS vendors, had installations targeted by the DIR deployment. The final aim of the project was to realize medical imaging data sharing through XDS-I, which would cover a wider affinity domain, including two other DIR’s. Altogether, those would completely cover the province of Quebec, with a total production of more than 10 million studies annually. The deployment is carried out in such a way as to allow pre XDS-I data sharing as an intermediary deliverable using a master patient index engine internal to the DIR solution, while insuring immediately a sufficient metadata richness to enable XDS-I publishing subsequently from the DIR which supports a combined imaging document source/repository actor. Vendors were required to model their interface according to a set of specifications published by the project group, and based as much as was possible upon standard IHE transactions.

 
Evaluation:
All RIS and PACS vendors having one or many products present on the sites to be integrated to the DIR were invited to prepare interfaces per the specifications. Vendors were met and the conformance of each vendor, with respect to the project specifications, was analyzed. When partial or non-compliance existed, it was examined to determine whether the interoperability gap could be filled in order to make the interface acceptable to the project.

The deployment of the edge server for the above-described jurisdiction was carried out successfully during the second half of 2009 with the RIS and PACS models most common across the geography of the project. This allowed verifying the adequacy of the specifications used by the project so other vendors could be given proven specifications. Considerable attention was dedicated to the issue of reliability in DICOM data transfer, as it was considered that the multiple interfaces involved increase the potential for anomalies, which could lead to imaging data loss through their purging from the producing site if they are improperly assumed to be transferred to DIR. The acquisition of many new PACS systems was carried out during the course of the project, but such newer systems were not equipped with long-time archiving capabilities as it was assumed that the DIR would to carry out that task. The go-live of such PACS systems had generally to occur before the commissioning of the DIR interface. Therefore, the full commission of the interface of such PACS had to be followed by a migration to DIR of HL7 and DICOM data acquired between the PACS go-live and the commissioning of the DIR interface. Again, considerable attention was paid to quality assurance in the archiving of migrated data before flagging the data as being eligible for purging.

 
Discussion:
The specifications of the project for building an interface to the DIR were successfully implemented, and allowed for the commissioning of an edge server for local systems to use to interface to DIR. During the course of 2010, different RIS and PACS vendors will be added to the DIR. Although any subsequent connection to the DIR of a RIS of PACS system is easier than the first deployment for that system, the necessity to adapt to each local ADT flow, which tend to differ between different healthcare enterprises, involved a minimum amount of work even when no new system was introduced. The project made no attempt to promote the normalization of ADT transactions as most local systems, imaging or other, have typically strong dependencies on the ADT configuration. It would therefore be unrealistic for a domain specific project to impact the IT organization of participating sites to that extent, so that this was left out of the scope of the project.
 
Conclusion:
The deployment and integration with local sites of a DIR was successfully demonstrated. It was accomplished using an edge server based technology, interfaced to local systems using specifications devised by the DIR project group. The solution was used to enable, as an intermediary project deliverable, a DIR-wide longitudinal view of imaging data. Nevertheless, it was possible to carry out the initial deployment while insuring that the solution would lend itself to the subsequent introduction of the IHE XDS-I profile, to which other DIR’s will participate, thus constituting a wider affinity domain.
 
References:
IHE Radiology TF, http://www.ihe.net/Technical_Framework/index.cfm#radiology