Speaker
Description
URL for further information
www.grycap.upv.es
Impact
Most European healthcare centres use today the DICOM Standard for storing medical images, but only a very few uses structured reports to code diagnostics.
Actual projects dealing with DICOM image processing (MDM, MEDICUS, TRENCADIS) use the Grid to store DICOM images and use metadata databases for coding the demographic information associated. Furthermore, TRENCADIS use DICOM-SR to code diagnostics through ad-hoc components, which could be improved by using existing components such as AMGA. The management of structured reports increases their potential and the use of AMGA provides a standardized mechanism in a Grid infrastructure.
The use of XML Schemes will also enhance browsing the information by different users. AMGA coding is much more natural for tree-organized data structured, which can be updated defining more information from the DICOM-SR files to the searchable metadata in AMGA, with the infrastructure benefit of replication, VO management and efficient access of AMGA.
Conclusions and Future Work
The work introduces an approach to organise data from DICOM Structured Reports and to enable hierarchical and fast searches using selected parts of its metadata.
The work is developing a tool that organises data integrated with a metadata catalog (AMGA), which seamlessly integrates with other gLite components as VOMS, SEs, LFC and processing through the WMS and CEs.
The work shows that the performance of the system is reasonable for research and training activities.
Detailed analysis
This work describes a tool that organizes DICOM-SR data by means of AMGA.
The contents of DICOM-SR can be seen as a data tree in which each leaf is one report element. Report elements can be of different types, each one conceptualized as a code (concept name) that unequivocally identifies the pieces of data. Different element types have different values.
All DICOM-SR reports are associated to a template. Each one corresponds to a given diagnostic evaluation (e.g. lung cancer screening) and determines the structure of any report. The templates are coded in a XML well-defined language.
Before inserting data from DICOM-SR report inside AMGA, all templates are built in AMGA as a DICOM-SR tree. Leafs of the tree corresponds to an AMGA directory identified by its “concept name” and each directory stores the values and references to the DICOM-SR report.
The tool creates automatically directory trees on the AMGA, inserting the data from DICOM-SR reports and enabling searching on them.
Keywords
metadata, DICOM-SR, AMGA
