Development Themes
To create an environment independent standardized CAT installation all software components of the imaging pathway (see Fig1.) can be converted to Docker containers. The following containers will be created (or tested if already available).
- Clinical Trial Processor (http://mircwiki.rsna.org/index.php?title=MIRC_CTP)
- PostgreSQL CAT Key-Database
- DCM4CHE PACS (http://www.dcm4che.org/)
- MAASTRO MIA (https://github.com/maastroclinic/)
- SeDI (http://semantic-dicom.org/)
- Blazegraph (https://www.blazegraph.com/)
After all components are Docker-ized they need to be collected into a network/vlan of containers to enable communication and data transfer between containers. It should be possible to build the entire pathway in one Docker-compose to be re-used.
The Common Workflow Language enables could provide a standardized framework to run our data mining programs/scripts/etc. Using Pentaho containers and CWL build a standardized data mining workflow. http://www.commonwl.org/draft-3/UserGuide.html
To limit the local hardware required to build CAT connectors for the network, moving the image analysis to the cloud is an interesting use-case. To do this we need to modify the MIA framework in a way that all the micro-services can run at a centralized location with a local file and result service. The biggest challenge is changing the current universal worker module to use JSON input instead of the DICOM files.
Fig1. Overview of the to-be-developed infrastructure