Ringholm-Logo Ringholm  Whitepaper Ringholm page header

Structured Reporting, a Powerful DICOM mechanism

The contents of this whitepaper are public domain.
See http://www.ringholm.com/docs/02030_en.htm for the latest version of this document.
Author: Herman Oosterwijk - president of OTech Inc. [OTech]
Document status: Final, version 1.0
Please sent questions and comments to herman.oosterwijk@ringholm.com


1. Introduction

Structured Reporting (SR) contains of what you might think of as leafs (aka content nodes) and relationships that can be configured in any way to construct a tree structure. The contents of the leaves are defined, for example they contain findings, measurements, codes, names, date/times, and, probably the most important, references to other DICOM objects.

Because of the DICOM references, a SR can be very tightly connected with DICOM images, waveforms, the description of the presentation of images (so-called presentation states), and any other DICOM objects you can think of. The relationships between the leaves are also specified and can be things such as �has properties�, �context�, constraints, concept modifiers, etc. Combined with the capability to develop a structure, SR is a powerful tool.

2. SR Implementation

There are certain constraints on the type of structure you can build. There are actually three different SR versions, each with increasing complexity, the exchange of which can be negotiated up front by the parties that want to exchange the information, and, as is common for every DICOM compliant device, as specified in their conformance statements. However, to make it easier for the recipient of these objects, there are also templates defined, which can either be used with the �generic� SR objects, or, defined it its own right for specific applications.

It seems unlikely that all diagnostic reports be replaced by this new DICOM SR. Even although it is possible to include the full text report as a �blob� of text in a SR, it would not make use of the power and capabilities of SR. Currently, there is a relatively good infrastructure in the HL7 world with long established products that seem to deal with this information rather well. However, there are some applications that could greatly benefit from the tight connection between the information in the SR�s and other DICOM objects such as images, especially when using the restricted templates. These area�s are:

  • Computer Aided Detection (CAD) output from a computer, e.g. what is used today to detect microcalsifications in mammograms , and what ever CAD applications will come along in the future (long nodule detection in chests, etc.). These objects will almost certainly be SR based. A structure is required when one needs a computer to generate output and SR certainly meets this need.
  • There are niches that today provide reports for a specific applications, such as Endoscopy, Ophthalmology, OB/GYN, etc. These applications are increasingly requested to integrate images as well. In the process of doing that, SR is a logical evolution to deal with images as well as text.
  • Cardiology generates procedure logs in a very structured format. These lend themselves very well to the SR application. Templates are being defined.
  • Ultrasound definitely needs a standard for their measurements. Many proprietary solutions exist that are in the process of being converted to SR.
  • The key image note to identify key images in a study might actually become the biggest driver of SR implementations, especially because image acquisition modalities are increasingly producing more and more images. New multislice CT scanners and MR acquisitions can do 100�s of images. The SR mechanism to provide image references with additional text providing the mage context (�significant images�) would allow identifying the selected images for a physician, not requiring downloading all of them.


3. SR and HL7 CDA

How will SR relate to HL7? It seems to be a logical fit in the Clinical Document Architecture (CDA). We need to still do some work. The Image and Integration SIG is mapping the DICOM information model to the version 3 RIM and we are mapping attributes. The encoding of the tree structure is actually done with recursive sequences in DICOM, something that could map relatively easily into an XML framework. We are far from being done, but progress is definitely being made. Who knows, hopefully we will do a combined SR/CDA demonstration at the next Integrating the Healthcare Enterprise (IHE) demo at HIMSS and RSNA.

4. References

[NEMA] "Digital Imaging and Communications in Medicine", http://medical.nema.org
[OTech] http://www.OTechimg.com/
[Clunie] "DICOM Structured Reporting", Dr. David A. Clunie, ISBN 0-9701369-0-0 http://www.pixelmed.com/


About Ringholm bv

Ringholm bv is a group of European experts in the field of messaging standards and systems integration in healthcare IT. We provide the industry's most advanced training courses and consulting on healthcare information exchange standards.