UNIGE.4

Grid Data Management

Long Title: Grid Data Management for Medical and Liquid Time Projection Chamber Image Analysis
Leading
Organization:
Université de Genève
Participating
Organizations:
Universität Bern
Domain: Grid
Status: finished
Start Date: 01.01.2010
End Date: 30.06.2011
Project Leader: H. Müller

This project has created a usable platform for research and collaboration using the Grid. Safe storage for medical images and metadata, definition of privacy levels, secure access and provision of Liquid Time Projection Chamber (LTPC) 2D event images to scientists and students via Grid technology are main parts of this effort.


Results

  • A security framework for safe data storage and processing was implemented for medical image analysis.
  • Also the application for content-based image retrieval of physics images was developed, a technique never applied to these types of images of events.
  • A storage element was added to the Swiss distributed computing infrastructure (SMSCG) and can be used by the participating virtual organisations.
Component Description
Storage element web interface You need a Grid proxy certificate stored on a myProxy-server to access these resources
Endpoint Storage element endpoint (storage resource manager protocol)
Grid Data Management for Liquid Time Projection Chamber Data Analysis Contribution European Grid Infrastructure EGI user forum 2011, Vilnius, Lithuaniat
Safe Storage and Multi-modal Search for Medical Images Contribution MIE 2011 - XXIII International Conference of the European Federation for Medical Informatics, session Imaging and Biosignals, Oslo, Norway

The physics image retrieval system was developed to be much more complex than the initial system that was foreseen using GIFT. The installed system took also into account many of the comments of the physicists.

Initial Situation

Unfortunately, some state-of-the-art Grid technologies that enable e-Science collaborations live only through the community funding cycle. We have identified state of the art key components in Grid data management, that have the potential when properly integrated and deployed, to be used as a sustainable common service that can benefit the daily research being performed by the partner universities, potentially even more.


Goals

The novelty and contribution of our approach applies to two areas:

  1. creating a Grid service for analyzing medical images based on the specifications found in the patient meta data, stored in the DICOM (Digital Imaging and Communications in Medicine) format and
  2. creating a Grid storage of Liquid Time Projection Chamber (LTPC) 2D event images and finding similar events by using image analysis software (GIFT - GNU Image Finding Tool) and event meta data, so the same technologies that are being used in the medical imaging use case.

The usable platform for research and collaboration using the Grid and GIFT will integrate:

  • a functional prototype of a safe storage system for medical images and their meta data, integrated with a PACS system, and an on-demand image analysis system.
  • privacy preservation on medical databases: privacy levels can be defined on a per-user basis. When a researcher searches images by similarity, only the ones matching his privacy level are shown.
  • digital certificate-enforced secure access to medical images. The technology can later be expanded to combine images from several institutes (only anonymized data will be used for the prototype).
  • LTPC event images provision to scientists and students via Grid technology including security, and offering automatic pattern recognition for pre-selection of different physical processes.

For the physics research community in Berne, the long-term effect of the project will be to set up a tool for daily operations (Grid storage).

Steps

The phases of the project are as follows:

  1. Installation of the required hardware and software in the hospital will be performed, including a VOMS server, MDM, DPM, AMGA and 3 instances of the Hydra service.
  2. Installation of all the required hardware and software at LHEP will be done. This includes DPM, GIFT and possibly AMGA (the SMSCG VOMS server will be used).
  3. Images will be provided on the storage in the sites.
  4. Prototype of successful communication in meta data search among medical and LTPC images will be delivered.
  5. Prototype of successful communication between PACS, GIFT and MDM will be demonstrated.
  6. A user interface will be developed. This will be browser-based to ease the employment.
  7. The software products of the project will be released with an open source license respecting the original licenses used in the current state-of-the art software.

Back