Problem Solving Environments for Medical Image Analysis Applications

During their lifecycle, Medical Image Analysis (MIA) applications go through several phases, namely development, parameter optimization, evaluation and clinical deployment. Each involves different tasks that are performed by users playing different roles and using systems with complementary functionality. In practice, these systems operate in stand-alone fashion, and the users involved are not always aware of each other. In an ideal scenario, systems and users would interoperate and cooperate in an integrated problem solving environment (PSE).



Figure 1: Information flow across phases in the lifecycle of a MIA application.

This project aims at the construction of a PSE to support MIA applications with three systems (DeVIDE, Nimrod and AMC-DWMS) and mechanisms to facilitate information flow across phases of the lifecycle (see figure 2). The PSE is developed in the scope of the ``Medical Diagnosis and Imaging'' subprogram of the Virtual Laboratory for e-Sciences Project , and adopts infrastructure available at the VL-e Proof-of-Concept environment .



Figure 2: Architecture of the proposed PSE.

DeVIDE, or the Delft Visualization and Image Processing Development Environment, is a platform-independent, interactive system that facilitates the rapid implementation and testing of component-based applications. DeVIDE has two main modes of operation: In interactive mode, it facilitates run-time steering of network execution via a GUI. In command-line mode, saved networks can be restored and configured to be executed without the GUI. In this way, complete networks can be visually constructed and explored, and later used as black-box third-party components in other applications. DeVIDE is used interactively during the development phase, and in command-line mode in the subsequent phases.

Nimrod facilitates the management of large experiments on distributed resources. The experiment is presented to Nimrod in a simple ``plan'' file describing the files to be transferred to the execution node, the tasks to be executed, and the files to be copied back to the source node. Nimrod is a valuable tool for parameter sweeps in the parameter optimization phase, and for image sweeps in during the evaluation phase.

The AMC-DWMS, or the Academic Medical Center Distributed Workflow Management System, performs the deployment of the custom MIA applications in a desktop grid inside the hospital. The data logistics and analysis is automatically performed by AMC-DWMS based on pre-defined workflows programmed in XML. The system supports a large variety of tasks, such as DICOM image import-export, caching, analysis and notification services.

Current Status

The first prototype implementation has been completed. All systems (DeVIDE, Nimrod and the AMC-DWMS) can successfully run a simple MIA application in an experimental environment. Information is saved in log files that can be later retrieved to restore a particular situation that generated a problem when the DeVIDE network was run in command-line mode from Nimrod or AMC-DWMS.

For more info see the VL-e Medical GFORGE page.

Team

This project is carried out by a large multidisciplinary team of researchers associated with several Dutch institutions:

Publications

K. C.Maheshwari, S. D. Olabarriaga, C. P. Botha, J. G Snel, J. Alkemade and A. Belloum. "Problem Solving Environment for Medical Image Analysis" IEEE Intl. Symp. on Computer-Based Medical Systems (CBMS 2007), Maribor, Slovenia, June 2007
abstract pdf

S.D.Olabarriaga, J.G. Snel, C.P. Botha, R.G. Belleman, "Integrated Support for Medical Image Analysis Methods: from Development to Clinical Application", IEEE Transactions on Information Technology in Biomedicine, Jan 2007
abstract pdf (restricted access)

Contact

Silvia Delgado Olabarriaga, PhD
Academic Medical Center (+31 20 566 8977)
Institute of Informatics (+31 20 525 7549)
University of Amsterdam, The Netherlands

silvia -at- science -dot- uva -dot- nl
S -dot- D -dot- Olabarriaga -at- amc -dot- uva -dot- nl

See more information here.

Acknowledgment

This work is carried out in the context of the VLe project, which is supported by a BSIK grant from the Dutch Ministry of Education, Culture and Science (OC&W) and is part of the ICT innovation program of the Ministry of Economic Affairs (EZ).