Vre ci presentation -ric workshop - july 26thPresentation Transcript
A Virtual Research Environment for Cancer Imaging (VRE-CI) RIC Workshop 26 July 2010
Outline Project Overview Background VRE-CI Progress Current activities Future Work Questions
A Virtual Research Environment for Cancer Imaging (VRE-CI) VRE-CIproject is funded by the Joint Information Systems Committee (JISC) to provide a framework to allow researchers and clinicians involved in Cancer Imaging to share information, images and algorithms.
JISC VRE frameworks phase 3.
22 months 01/05/2009 – 28/02/2011.
Project Partner: Microsoft Research
PI. Prof. Anne E. Trefethen
Co-I. Dr. Vicente Grau
Project Manager Dr. M. Susana Avila-Garcia.
Xin Xiong, Charles Crichton and Andrew Tsui.
Implementing algorithms developed in the past by other researchers.
Sharing is an issue since they commonly use different programming tools, data sources and formats.
Reusing existing pieces of code.
? They work at different locations Computing Lab
Researchers show resultson a computer/laptop screen.
Clinician do not have the opportunity to interact with the applications.
No user interface defined for clinicians.
Virtual Research Environment Sharing algorithm to:
Reuse it as it is.
Reuse it as part of a bigger solution.
End users Research Context Code Share associated objects:
Original data sets.
Web Services Test the algorithm:
Different image formats
Scientific Workflows/ Activities
Open discussion forums.
Scientific Workflows/ Trident
This graphical representation hides the complexity of the applications allowing to:
understand at a high level how the application works,
modify the flow of activities in an easy way and
analyse the application into the level of required detail
VRE-CI Extend the functionality of the Research Information Centre (RIC) and include tools for articulating and sharing imaging algorithms through the integration of Trident.
Oxford University Groupware Oxford University has adopted a Microsoft solution as the integrated framework to support future communications and collaboration across the University. This project will provide an exemplar within the realm of the Virtual Research Environment and will result, we hope in a model that can be utilised in other parts of the University.
Oncological image analysis
Researchers at Oxford University, Churchill Hospital and Microsoft Research Cambridge. Novel segmentation algorithms for colorectal cancer images.
Define research objects for the project and make them available to the team members.
Gray Institute for Radiation, Oncology and Biology.
Ensure the system is interoperable with existing image management systems.
Oncological image analysis Developments focus on: Image analysis: Define web services and web parts. Image functionality: Load Images DICOM, regular image formats. Search on image metadata. Image viewer.
Image analysis - Results
Image metadata SharePoint Lists:
Search on image metadata Search on image metadata. Modality, Study Description and Keyword Keyword. (Modality, Anatomical part, Manufacturer, Place, Person, Date, Disease).
Search Operation: And, Or
Image Viewer Adapt an existing Deep Zoom viewer [http://images.cancergrid.org/score/] Need a DICOM image filter to convert to bmp or jpeg format. Developed using the Insight Toolkit ITK.
What is Silverlight DeepZoom? Multiple layers at different resolutions Enables quick zooming and panning of large images Requires Silverlight browser plugin
Trident Integration in RIC Demo developed in Microsoft Research: Kindly demonstrated by project manager Oscar Naim. Final deployment will require SharePoint 2010. Demo includes functionality to perform and track workflow executions. Future developments will include the workflow composer => Highly important for our project. At OeRC Tests to install RIC on SharePoint 2010. By defining our image processing as web-services we will easily be able to integrate them when they arrive
Summary Project update: Web service and web part defined for Level set algorithm. To be integrated as workflow activity. Information model definition for regular formats and MR, CT, US and PET DICOM images. Advanced search functionality Adaptation of Image viewer. Current activities: Installing the RIC on SharePoint 2010. Integrating image viewer and search functionality Feedback is being collected from the user pilot group to improve and enhance the system before making it available to a wider imaging research community.
Future work Image-based functionality Anonymise DICOM files uploaded into the system. Include functionality for additional formats: Analyze, Mhd images ROB image formats . Image viewer integration.
Remote Image Access
Create link to original datasets and keep low resolution copy at the server.
Trident integration: Test Trident demo application in our setup. Define catalogue of activities for Trident.
Acknowledgements British Library – Stephen Andrews Niranjan Joshi Interviewees at Gray Institute of Radiation Oncology and Biology. Sean Smart, Danny Allen