Closing Workshop of the VL-e Medical Project

The closing workshop of the Medical Diagnosis and Imaging (VL-e Medical) subprogram of VL-e took place at the AMC on December 16, 2009. The workshop was co-organized by VL-e and GridForum Nederland.

Program

See also the summary of talks and the description of demos

Summary of Talks

Opening

Silvia Delgado Olabarriaga, Bioinformatics Laboratory, AMC

This talk presents a brief overview of the VL-e Medical project (partners, collaborators, activities) and introduces the workshop program (talks, demos, closing discussion).

Baby Steps in VL-e Medical

Robert Belleman, Informatics Institute, UvA

If anything, the VL-e project was about making bridges. For the six application subprograms in VL-e, the bridge that had to be built for Medical Diagnosis and Imaging was probably the most challenging. In this talk I take you on a short trip down Memory Lane and look back on this project: what was planned, what worked, what didn't and what have we learned?

Grid and cluster computing in VUmc; Future prospects and the role of VL-e medical

Bob van Dijk, VUmc

At the VUmc the vl-e medical program has focussed on brain imaging research. In my talk I will describe the typical neuroscience researcher and his/her needs for computing resources. I will describe our attempt to bridge the gap between those needs and the existing GRID resources. Next I will describe the first results of the neuGRID program; that aims to bring an infrastructure that will bring AD MRI research data, to AD MRI researchers; using GRID technology.

Overview and remaining challenges of medical imaging on the 3.0 Tesla MRI facility at the AMC

Aart Nederveeen, Radiology AMC

In this talk the VLe-the activities in the AMC will be reviewed. Strength and weaknesses will be identified. Implementing an IT infrastructure that benefits clinical routine has been proven difficult in the past years. Still several challenges remain and will be summarized.

Distributed User Support for Troubleshooting

Eduard Drenth, Logica

At this closing workshop I will present our work on introducing agent technology in grid research. Together with the AMC we were able to develop a promising pilot that demonstrates the possibilities agent technology offers in a multi domain and distributed environment like the grid. I will point out that follow-up will lead to improved usability, stability and cost reduction.

Fault-tolerant nested workflows for dual tensor atlas estimation

Matthan Caan, AMC Radiology and Bioinformatics Laboratory, TU Delft

Research over the past four years has resulted in a method for
estimating a dual tensor atlas out of Diffusion Weighted MRI data of the
human brain. This atlas describes up to two white matter bundles per
voxel. The atlas is estimated from a group of 100 healthy subjects. In
the first workflow all pairs of subjects are to be coregistered, i.e.
100*100 jobs. The second workflow defines the average atlas and warps
all subjects. In the third workflow the dual tensor model is estimated
per voxel.
We developed a dedicated nested workflow containing the aforementioned
MOTEUR-workflows. Our framework is fault tolerant, i.e. erroneous jobs
are identified and resubmitted. A dedicated front-end is developed to
grant easy access to end-users.

VL-e-med-related medical imaging experiments at CNRS-Creatis

Tristan Glatard (CNRS Lyon, Creatis)

Software, experiments and experience gathered within the VL-e medical project open important perspectives regarding grid exploitation in medical imaging. At Creatis (CNRS, France), we are following this trajectory by building up an execution environment for distributed applications based on components integrated by VL-e med. This talk will present on-going grid experiments carried out at Creatis. In particular, workflows related to cardiac MRI segmentation/evaluation, radiotherapy simulation and ultrasound simulation will be presented. For each experiment, motivations, implementation and early results will be discussed, illustrating how VL-e medical results are used in a more general context. Finally, some remaining challenges will be highlighted.

Will VL-e Medical boost e-life-sciences?

Antoine van Kampen (Bioinformatics Laboratory, AMC)

e-science approaches increasingly penetrate life sciences research as this potentially provides new and better approaches towards collaboration, information management and computing. In the Netherlands, the NBIC BioAssist programme aims to use e-science technology to setup bioinformatics support. In the AMC we recently initiated a e-science research group that aims to develop applications for medical imaging and for life sciences. E-science for life sciences (i.e. e-biosciences) is progressing slowly for various reasons. This raises the question: What can we learn and use from VL-e Medical to accelerate progress in life sciences? In an attempt to answer this question I will present a brief overview of the BioAssist programme and the AMC e-bioscience group, show few examples of bottlenecks we encounter in e-bioscience and, subsequently, give my personal view about the future of e-bioscience.

NeuroLOG* *: a collaborative environment for neurosciences interfaced to the EGEE production grid

Johan Montagnat (CNRS Sophia Antipolis, FR)

Neurosciences involve large-scale studies over ever growing data sets populations. Although federating neuroscience resources, such as population of data and neuro-image analysis tools, has become increasingly needed, the set up of multi-centric studies remains challenging due to the heterogeneity of data sources, the complexity of data models, the specificity of image analysis tools and the sensitivity of neurological data. Leveraging Grid technologies, the NeuroLOG projet (http://neurolog.polytech.unice.fr) is designing and implementing a secure and distributed collaborative environment for neurosciences. It enables the federation of heterogeneous neuro-data stores using data mediation, knowledge representation and data federation technologies. It provides support for neurosciences data analysis pipelines design and distributed execution. It delivers high performance by integrating resources of the EGEE European production grid infrastructure into the usual working environment of neuroscientists.

Process Modeling and Workflow Management for Medical Research in the German D-Grid Initiative

Andreas Hoheisel (Fraunhofer Institute, Berlin, DE)

Executing complex processes in heterogeneous distributed computing environments is a major goal in applying Information and Communication Technologies in medical sciences and clinical applications. Within several projects of the German D-Grid Initiative we enhanced the "Grid Workflow Execution Service" to simplify the process modeling and to speed up the invocation of IT processes on distributed resources. The talk focuses on recent challenges regarding security, fault-tolerance, and scalability of workflow approaches in the health care sector.

Gridbased Medical Image Processing in the German* MediGrid Achievements and Challenges

Dagmar Krefting (Charite Hospital, Berlin, DE)

n 2005, the German national Grid initiative D-Grid has launched the community grid for life-sciences and medicine, the [[http://www.medigrid.de/][http://www.medigrid.de][MediGrid]] project. To date, different medical imaging and biosignal processing applications are implemented, encompassing the fields of Neuroscience, sleep-medicine and hemodynamics. With respect to "classical" grid communities as high energy physics or astrophysics, additional challenges in usability and security appeared. Solutions, approaches and open issues in the exploitation of the Grid infrastructure for medical image processing are presented and discussed

Demos

Grid Workflow Efficient eNactment for Data Intensive Applications

Johan Montagnat, I3S CNRS Sophia Antopolis, FR
Tristan Glatard, Creatis CNRS Lyon, FR

e-BioScience infrastructure for sequence analysis

Angela Luyf and Barbera van Schaik, Bioinformatics Laboratory AMC, NL

MEDIGRID

Dagmar Krefting, Charite Hospital Berlin, DE
Andreas Hoheisel, Fraunhoffer Institute Berlin, DE

Distributed User Support for Troubleshooting

Eduard Drenth and Wico Mulder, Logica, NL
See related talk

DTI atlas service

Matthan Caan and Carsten Byrman, Bioinformatics Laboratory and Radiology AMC, NL

During the demo we can monitor the progress of the workflow, and inspect a previously calculated atlas via interactive fiber tracking. The setup of the workflow, as well as other aspects not mentioned in the talk can be discussed here. See related talk

ContactUs