You won’t find any of the apps developed at the SIIM Hackathon available at the iTunes store, but the event provided a great test kitchen for shiny, new, Web-friendly versions of DICOM and HL7 that promise to speed the development of applications for medical imaging while preserving a DICOM standard-based backend.
At the start of the recent SIIM meeting in Long Beach, a group of hackers comprised of vendor programmers, expert coders (dubbed Code Heroes), residents and radiologists occupied a room equipped with tables, chairs, energy drinks and WIFI. From there, they accessed recently released DICOM Web APIs; a new HL-7 REST API from FHIR; patient data and image access APIs from McKesson and Agfa; and an assortment of miscellaneous radiology resources, all uploaded to the cloud.
From this brew, emerged a handful of applications developed in just three days, many capable of accessing multiple databases to create a single view of the desired data, all safely over the Internet. “This is probably the most important thing that has happened in our world since DICOM has come onto the scene,” Don Dennison, consultant and Hackethon organizer, said of the new web-friendly additions to DICOM and HL-7. “Image sharing and including images in the EMR are the two areas that are going to benefit the most.”
DICOM is notoriously cumbersome and difficult to learn, so the introduction of DICOM Web will not only speed development in healthcare, it will broaden the pool of programming talent. “We now can unlock the skills and talents of the people doing web apps; and there is an army of people out there,” Dennison adds. “We are now ready to work the way they want to work.”
The first three APIs were released in October 2013: QIDO-RS (query, lookup), WADO-RS (retrieve), and STOW-RS (storage). The RS refers to REST-ful services, REST being a software architectural style that simplifies machine-to-machine Internet communications and is widely deployed in web services.
Its counterpart on the HL-7 side, FHIR, is available from HL-7 as a Draft Standard for Trial Use, and was initiated several years ago in response to criticism that HL-7 was hampering interoperability. FHIR, which stands for Fast Healthcare Interoperable Resource, also is based on REST-ful web services, and incorporates basic Hypertext Transfer Protocol (HTTP) operations such as Create, Read, Update and Delete, or CRUD.
Building on Existing Applications
At 3PM on Saturday, May 19, Dennison called the individual hackers to the front of the room to present their work. Mark Kholi, MD, Indiana University, described himself as “a radiologist who knows enough programming to be dangerous.”
“It was really fun to come in here, and, with very little effort, grab a patient’s report out of a repository, create a new patient report and push it back into the repository,” he said, then see the report show up in applications others were creating. “And they were able to consume that report and mix and match it with other things.”
Instead of building an application from the new APIs, a team of developers from Dell (Brian and Dave) took an existing feature set in their VNA and integrated the APIs into Dell’s core solution.
Their focus was a “legacy” DICOM feature of the archive called query spanning service, whereby a request is broadcast to various DICOM repositories and the studies that are returned are merged to present a universal DICOM view of the data. The pair aimed to make this feature work by merging test studies from the Dell archive with studies from the McKesson archive, the latter accessed using the new DICOM Web QIDO.
“Here is your typical workstation viewer running on windows, there is a list of studies, three of which are coming from our archive and three of which are coming from McKesson,” Dave said—to a round of applause.
Translating the achievement into real-world terms, Dennison compared the activity to someone sitting at a PACS-workstation client on a local system who wanted to check against the regional HIE but lacked a LAN or VPN connection. Using the app, that someone would go over the Internet using HTTPS, using the local context, collect data into a complete longitudinal patient record and present it as if in one system. “REST-based APIs allow you to traverse these things and not just patient record data,” Dennison says. “As we heard about HL-7 FHIR, it also can be bottomless; it can be anything.”
A Couple of Days of Effort
Dave Reich from