DICOM PS3.17 2024a - Explanatory Information

YYYY.5.3 Research

While imaging data may be important for research activities, it is rarely used solely by itself. It is generally used in conjunction with other aspects of the patient medical record - diagnoses, treatments, outcomes. Thus, support for imaging related research needs to support integrated activities with other healthcare informatics systems and data.

Research functions must also not impact ongoing healthcare operations. Data for research is therefore typically extracted from clinical operational systems and transferred to a separate server, often with patient de-identification. These systems are sometimes denoted a "data warehouse", an extract of operational data that can be sorted, filtered, and analyzed in any number of ways to support research questions.

The Inventory might thus support research use cases in several ways. In the broadest sense, since it is a representation of the imaging repository database, it can be used for imaging research in conjunction with the image instances and the medical record data. As a DICOM object, it can be transferred to other systems for further processing. Since the data is in a standard format, it can be processed using readily available tools without having to know the proprietary table layouts of theimage management system database. And as the Inventory has links to the stored SOP Instances, further drill down to the image instances and more detailed metadata is facilitated.

A complete Inventory might be used for research purposes, especially if it has already been extracted for other purposes (such as safety backup). Such an Inventory may have its data transformed, de-identified, and loaded to a data warehouse. But a more focused Inventory might be produced for specific research processes. In particular, if searches of an EMR or data warehouse produces a census of candidate Studies, an Inventory of for just those Studies may be created using List of UID matching on Study UID in the Scope of Inventory, and the Inventory content could be further constrained by other Attributes. It should be noted, however, that the filtering of Studies by the defined Scope of Inventory is not sufficient for most research purposes, but it may be sufficient as a first level selection that simplifies additional filtering by other processes.

In most research uses, data sets must be de-identified. However, as the Inventory must typically be linked (via Patient ID) to other patient medical records, care must be taken in processing of Inventories for research to ensure de-identification. The approaches will vary depending on the specific research questions and data used, and the overall medical record architecture and systems of the organization. See Section YYYY.6.8.

The Inventory IOD is defined with the data elements necessary to support the primary use case of migration. However, the image management system may manage additional Attributes at the Study, Series, or Instance levels that might be beneficial for research, and that could be included in the Inventory as Standard Extended Conformance.

DICOM PS3.17 2024a - Explanatory Information