Experience implementing a DICOM 3.0 multivendor teleradiology network.

OBJECTIVE

The ISIS Center at Georgetown University received a grant from the U.S. Army to act as systems integrator for a project to design, develop, and implement a commercial off-the-shelf teleradiology system to support the U.S. troops in Bosnia-Herzegovina. The goal of the project was to minimize troop movement while providing primary diagnosis to military personnel. This paper focuses on Digital Imaging Communications in Medicine (DICOM) 3.0 related issues that arose from this type of teleradiology implementation. The objective is to show that using the DICOM standard provides a good starting point for systems integration but is not a plug-and-play operation.

METHODS

Systems were purchased that were based on the DICOM 3.0 standard. The modalities implemented in this effort were computed radiography (CR), computed tomography (CT), film digitization (FD), and ultrasonography (US). Dry laser printing and multiple-display workstations were critical components of this network. The modalities and output devices were integrated using the DICOM 3.0 standard. All image acquisition from the modalities is directly to a workstation. The workstation distributes the images to other local and remote workstations, to the dry laser printer, and to other vendors' workstations using the DICOM 3.0 standard. All systems were integrated and tested prior to deployment or purchase. Local and wide area networking were also tested prior to implementation of the deployable radiology network.

RESULTS

The results of the integration of the multivendor network were positive. Eventually, all vendors' systems did communicate. Software configuration and operational changes were made to many systems in order to facilitate this communication. Often, software fixes or patches were provided by a vendor to modify their DICOM 3.0 implementation to allow better communications with another vendor's system. All systems were commercially available, and any modifications or changes provided became part of the vendor's commercially available package.

CONCLUSION

Seven DICOM interfaces were implemented for this project, and none was achieved without modification of configuration files, changes or patches in vendor software, or operational changes. Some of the problems encountered included missing or ignored required data elements, padding of data values, unique study identifiers (UID), and the use of application entity titles. The difficulties with multivendor connectivity lie in the understanding and interpretation of standards such as DICOM 3.0. The success of this network proves that these problems can be overcome and a clinically successful network implemented utilizing multiple vendors' systems.