1. MITK Diffusion (under the hood)
Klaus H. Maier-Hein (Ph.D.)
Computational Disease Analysis Group
Medical and Biological Informatics

3. Reproducibility
 anything less than the release of source programs is intolerable for results that depend on computation

4. Different levels of interoperability
Different user-groups use and extend our software
Department
Research Center (Physics, Radiology)
„Sonderforschungsbereich“ (special research area)
Collaborators / “the world”
MITK supports data- and code interoperability
Code: “component-based” approaches
Data: Nifty/Nrrd, DICOM (Q/R), XNAT, (NITRC) …

5. Data interoperability
Different Interfaces
REST, Webinterface, -applets, DICOM q/r
Very flexible
Upload, view, search, meta-data, manage access
Support by Slicer, NiftyView, CTK (soon available)
Data infrastructure of choice for SFB Heidelberg
Centralized data management resource to support multiple investigators and research studies

6. Code interoperability
Scientists don’t “waste” time on software architecture
Requirements are still demanding
Re-usability
Interoperability
Collaboration
Clinical application
MITK uses component technology
to minimize the effort

8. The BlueBerry Application Framework

10. The MITK Workbench
Views
Editors

11. The BlueBerry application framework
What is BlueBerry?
The BlueBerry application framework
Similar to the Eclipse RCP (but in C++)
Builds plugin-based applications (OSGi-driven)
Lazy loading of plug-ins (scalability)
Strong encapsulation & loose coupling
Reuse/recombine already existing plug-ins
Flexible application layout
„views“ and „perspectives“
Written for and included in MITK (but independent)

12. CTK Plugin Framework

13. Application Hosting (DICOM WG23)
CTK
CTK is an international initiative to provide a shared code base for re-usable components and interoperability technology.
Plugin Framework
Command Line Modules
Widgets
DICOM support
Application Hosting (DICOM WG23)

14. CTK Plugin Framework

15. Dynamic OSGi-based Framework
CTK Plugin Framework
Dynamic OSGi-based Framework
Enables service oriented architectures
Enables distributed/large-scale applications

16. C++ implementation of the OSGi Service Layer:
CTK Service Registriy
C++ implementation of the OSGi Service Layer:
Service registry
Publish
Find
Service description
Service provider
Service requester
Bind
API is very close to the OSGi Service Layer specs

17. Plug-in structure Unique name Meta information Third-party
dependencies

18. Simple and non-intrusive
Plugin Framework
Simple and non-intrusive
The API is surprisingly simple, services require no special interface
Reduced complexity
Focus on interface/service, internals hidden
Versioning
Plug-ins and their dependencies are versioned

19. Command Line Interface

21. Command Line Interface
Integrating command line programs
Executables use XML to describe their input and output
Communication via standard input/output channels
Examples
ITK based registration programs
Compiled Matlab code
Slicer CLI modules
Command
arguments

22. MITK – modular application development at different levels: Toolkit:
Summary
MITK – modular application development at different levels:
Toolkit:
Use MITK shared libs + µServices
Application Framework:
Create dynamic and extensible applications using BlueBerry and CTK
MITK Workbench:
An extensible end-user application for rapid prototyping

23. MITK (Diffusion) is a lot more than just a bunch of algorithms
Conclusion
MITK (Diffusion) is a lot more than just a bunch of algorithms
CLI Modules supported by Slicer, GIMIAS, MITK, (MedInria)
Very easy to adapt to the standard
Immediate visibility through widely applied platforms
CTK and BlueBerry allow rich plugin-based applications
Everybody profits from developments
Nobody wants to repeat implementations of e.g. DICOM (Diffusion) or XNAT interfaces
Enables and eases distribution and reproducibility of research results

24. Medical and Biological Informatics German Cancer Research Center

25. MITK & CTK Plattform group

26. Computational Disease Analysis Group
Jonas
Michael

27. Thank you!

28. Implemented OSGi specifications in CTK
Plugin Framework
Implemented OSGi specifications in CTK
Log Service Specification
Provides a general purpose message logger.
Metatype Service Specification
Provides a unified way to describe metadata about services.
Configuration Admin Service Specification
Allows to set the configuration information of deployed plugins.
Event Admin Service Specification
Inter-plugin communication mechanism based on a event publish and subscribe model.

29. OSGi Specifications OSGi Core Specifications are small
OSGi Service Compendium defines many optional services:

30. Plug-in structure Unique name Meta information Third-party
dependencies

31. manifest_headers.cmake
OSGi meta information
manifest_headers.cmake
set(Plugin-Name "A human readable plug-in name")
set(Plugin-Version "x.x.x")
set(Plugin-Vendor "A human readable vendor name")
set(Plugin-ContactAddress “Web page, , etc.")
set(Require-Plugin <list-of-plugin-symbolic-names>)

32. Increased complexity due to the dynamic nature of the system
Conclusion
Costs
Increased complexity due to the dynamic nature of the system
Services can come and go as they want
Components are tied to the BlueBerry application framework