UC-UPV-medical-images

From EGI Knowledge Base

Use Case title: Medical Image Analysis on the Grid.

Short description: Sharing and organising medical imaging knowledge is a key issue in medical research and training. Evidence-based medicine is also demanding high-quality well-organised knowledge bases to check for second opinion and drive diagnosis. However, sharing and organising medical imaging data is not straightforward. Technological and legal problems on exchanging data make it difficult or even impossible with the current infrastructures. On the other side, the index criteria used in clinical practice are inefficient when searching for knowledge.

Actors involved:

• Data Providers: Medical centres (mainly hospitals) with Image Diagnosis Departments.
• Users: Radiologists.
• Application Developers: Integrators of the medical imaging processing tools, data interfaces and side-applications.
• Operators: In charge of the administration of the users, the update of templates and the maintenance of data gateways, central services and processing resources.

Related Requirements: The requirements are analyzed in four areas:

• Computing. Depends very much on the number of users and cases involved. A figure of a sustained availability of 30-40 CPUs with peak demands of 50-75 will be reasonable for a case with 5-7 users.
• Storage. The storage requirements are low, since data is kept distributed at the local centres (in order to be compliant with the EU regulations). Data is temporarily copied in the main storage and local disks of computing services. Central services only store catalogues and user profiles.
• Security. User authentication and authorisation is managed through VOMS credentials. Services are “VOMS-aware” and only provide access if the appropriate credentials are presented. Data is encrypted to avoid malicious access from unauthorised users (even local users with administrative privileges at the processing services). Encryption keys.
• Deployment: The services must deal with the particularities of hospital environments. Some nodes must be deployed within the hospital network, which implies that firewalls restrict heavily the ports and protocols that can be used. Adaptation of GridFTP, for example, needs to be done.

Pre-Conditions: A hospital infrastructure must be set-up to deal with it.

Steps:

1. Producer
   1. Authenticate and validate group credentials.
   2. Select case and copy and register it in the global indexing system.
   3. Fill-in the metadata related to the diagnostic of the case.
   4. Submit the metadata.
2. Consumer
   1. Authenticate and validate group credentials.
   2. Select the ontology and fill-in the searching criteria.
   3. Select the post-processing action (if any).
   4. Submit the job (data retrieval to the processing services, processing and copy of the results).
   5. Retrieval of the results.

Post-conditions: A group of operators update the structured report templates and ontologies, used in the indexing of the cases, as well as the groups and roles of the users.

Projects involved: CVIMO, “Valencian Cyberinfrastructure for Medical Imaging in Oncology” (http://www.grycap.upv.es/cvimo). CVIMO is a platform developed to share and organise medical studies and reports based on ontologies constructed upon the fields of structured reports. It is based on a Grid Software Architecture of WSRF services that organise coding, access rights and data location for different studies and reports.

Middleware: The middleware used is a proprietary system called “TRENCADIS” (Towards a Grid Environment for Processing and Sharing DICOM Objects), which is based on WSRF and uses components from GT4 and gLite.

Application: Application for searching cases by content, writing structured radiology reports and processing images.