NOTE: As of November 2018, the new home of DICOM4QI is here:


1.Description of the platform/product:

  • name and version of the software: MeVisLab 3.0
  • free? yes
  • commercial? commercial licenses available on request
  • open source? no (only partially, sources included in above download)
  • what DICOM library do you use? custom (MLDicomTree, using some headers from dcmtk)

2.Description of the relevant features of the platform:

  • are both single and multiple segments supported? yes (reading), single-only (writing)
  • how are the overlapping segments handled? loaded separately (display depends on application)
  • do you support both BINARY and FRACTIONAL segmentation types? yes (reading), only binary supported for writing so far
  • do you render the segment using the color specified in the DICOM object? yes / possible (display depends on application)
  • how do you communicate segment semantics to the user? MeVisLab provides field interface, UI depends on application (no application available yet using this new API as of November 2016)
  • how do you support the user in defining the semantics of the object at the time segmentation is created? no support is available yet (support similiar to the QIICR web interface planned, but not yet implemented)

3.Read task: load each of the DICOM SEG datasets that accompany the imaging series into your platform

NB: MeVisLab is an IDE, not an end-user application. Therefore, both the rendering of the lesion masks, as well as the display of the meta information entirely depend on the actual clinical application prototype. The screenshots below are from a custom MeVisLab network quickly put together just for this demonstration.

The underlying improved reading capabilities become available to all clinical applications and prototypes built using recent MeVisLab versions. Work on integration into our internal "MeVisLab Application Framework", e.g. the extension of the object-relational mapping classes by SEG-relevant structures and the corresponding import code has begun, but it will take more work to produce nice frontends for end user-visible applications.

Reading-wise, the new module MultiFileVolumeListSEGOutput has been implemented (and will be shipped as part of the freely available FMEwork/ReleaseMeVis package together with MeVisLab), which offers not only the segmentation mask (for which the existing MultiFileVolumeListImageOutput is sufficient), but meta information about each selected segment. Again, this information is provided as MeVisLab fields, extending the internal API by the necessary pieces for application support; these components are not directly visible to the end-user.

Test dataset #1

Test dataset Result of rendering
3D Slicer

Test dataset #2

Test dataset Result of rendering
3D Slicer

Test dataset #3

Test dataset Result of rendering
3D Slicer

Test dataset #4

Test dataset Result of rendering
3D Slicer

4.Write task

  • segment the lung lesion using any method available in your platform; save the result as DICOM SEG; please include in the series description the name of your tool to simplify comparison tasks!
  • run dciodvfy DICOM validator; iterate on resolving the identified issues as necessary
    • support for SEG writing (in fact, support for writing multi-frame DICOM files) has been newly introduced in MeVisLab
    • no remaining issues reported by dciodvfy (except for Study attributes inherited from the input data)
  • run dcentvfy DICOM validator
    • PatientIdentityRemoved and DeidentificationMethod tags are not properly copied from input data