gms | German Medical Science

GMDS 2012: 57. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

16. - 20.09.2012, Braunschweig

A flexible virtual whole slide microscopy framework for distributed research

Meeting Abstract

  • Sven Friedl - Fraunhofer-Institut für Integrierte Schaltungen IIS, Erlangen, Deutschland
  • Carol Geppert - Pathologisches Institut des Universitätsklinikum Erlangen, Deutschland
  • Arndt Hartmann - Pathologisches Institut des Universitätsklinikum Erlangen, Deutschland
  • Friedrich Paulsen - Institut für Anatomie Lehrstuhl II der Friedrich-Alexander-Universität, Erlangen-Nürnberg, Erlangen, Deutschland
  • Thomas Wittenberg - Fraunhofer-Institut für Integrierte Schaltungen IIS, Erlangen, Deutschland

GMDS 2012. 57. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Braunschweig, 16.-20.09.2012. Düsseldorf: German Medical Science GMS Publishing House; 2012. Doc12gmds074

doi: 10.3205/12gmds074, urn:nbn:de:0183-12gmds0745

Veröffentlicht: 13. September 2012

© 2012 Friedl et al.
Dieser Artikel ist ein Open Access-Artikel und steht unter den Creative Commons Lizenzbedingungen (http://creativecommons.org/licenses/by-nc-nd/3.0/deed.de). Er darf vervielfältigt, verbreitet und öffentlich zugänglich gemacht werden, vorausgesetzt dass Autor und Quelle genannt werden.


Gliederung

Text

Introduction: New generations of whole slide microscopy scanning systems are approaching market. Bunches of slides can be scanned completely with high resolution in short times. Such whole slide images can play a major role in distributed multi-centre research projects. Probes, generated in one centre, can be scanned and digitally provided to the partners. Intelligent virtual slide frameworks can deal with the resulting large amount of data per slide by storing tiles of the slides in different resolutions and transferring them efficiently. However, most scanning systems are providing different proprietary data formats and current frameworks are either offered as remote storage services for one format or depend on a conversion to a general whole slide image format. Since promptness and flexibility are important factors for many distributed research projects, a delaying conversion or remote transfer of the data, as needed for commercial storage services, is often a hurdle.

Material and Methods: Previously, we presented a virtual slide framework for training purposes based on dzi images for efficient data handling [1]. This framework has been implemented using MS Silverlight technology to use standard web components. To assure a prompt distribution of newly scanned probes and to obtain full flexibility in data handling and processing, the framework technology is currently upgraded with an integrated on-the-fly tile generation. A third-party library is employed to deal with different proprietary data formats [2]. After a specific region of the digital slide is requested by the client, the framework generates these in the required resolution and provides it as needed by the standard technology. Tools to add comments and to annotate distinguished regions of the slide online are integrated as well. Thus, each research partner can deal with the probes and save own metadata with the slides on the server for exchange or further research. The integration of algorithms for advanced automatic analysis methods, as e.g. feature classification, is possible. The implementation with standard web technologies minimizes the requirements for the local IT-infrastructure.

Results: The developed framework will be introduced shortly in a distributed research project, dealing with automatic classification of histopathological probes. In an experimental setup, the efficiency has been tested prior to the routine usage. Measurements of the data transfer rates and computation loads on the server and the client have been evaluated. Even with the on-the-fly tile generation, an efficient transfer and visualization is observed. The client will recognize no difference in comparison to the converted data handling. However, the load of the CPU of the server is significantly higher. The transfer rates are comparable depending on the grade of optimization regarding intermediate image request during pan or zoom actions.

Discussion: The updated framework for virtual whole slide microscopy seems to be sufficient for distributed research projects, which rely on prompt and flexible data distribution. The absence of data conversions or transfers to remote servers offers an efficient approach with few requirements to the IT-infrastructure. For flexible or larger research groups, a scalable server infrastructure should be considered to prevent excessive server load.


References

1.
Friedl S, Oehmke G, Wittenberg T, Hartmann A, Geppert C. Freies Mikroskopieren im Web. In: 56. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (gmds), 6. Jahrestagung der Deutschen Gesellschaft für Epidemiologie (DGEpi); 2011 Sep 26-29; Mainz, Deutschland. p. 177-8.
2.
Goode A, Satyanarayanan M. A Vendor-Neutral Library and Viewer for Whole-Slide Images. Computer Science Department, Carnegie Mellon University, Technical Report CMU-CS-08-136. 2008;