Difference between revisions of "Developer Meetings/20170523"

From Slicer Wiki
Jump to: navigation, search
 
(4 intermediate revisions by the same user not shown)
Line 11: Line 11:
 
= To Discuss =
 
= To Discuss =
 
* Use of GDCM vs DCMTK in Slicer
 
* Use of GDCM vs DCMTK in Slicer
 +
* Definitive transition from mailing list to discourse
  
 
= Conclusion =
 
= Conclusion =
 +
* slicer.org on GitHub
 +
** It makes sense to store website on GitHub and have it served from Digital Ocean (or legacy server)
 +
** Waiting final transition to server is done, cronjob
 +
 +
* ITK: What ITK should create ?
 +
** Thin wrapper around DCMTK
 +
** Default behavior: if context is not known. Secondary capture seems to be the only valid approach.
 +
** Moving toward consolidating tool around DCMTK
 +
** Anonymization: David Clunie has implementation in Java. See http://www.dclunie.com/pixelmed/software/webstart/DicomCleanerUsage.html
 +
** [https://github.com/dgobbi/vtk-dicom/ vtk-dicom] from David Gobbi is nice. Issue is that it introduces dependency on VTK for tools that  wouldn't require it (e.g CLI) + VTK doesn't handle orientation (by default) in its core image data structure
 +
** DCMTK: currently support only to create secondary capture
 +
** DCMQI: support to create seg object
 +
 +
* Public repo with collection of DICOM
 +
** During last project week, discussed use of IPFS to store medical test data (DICOM, ...)
 +
** Stored at MIT, Bremen, Standford, ...
 +
** You can contribute without being the central server
 +
** It is easy to start  an IPFS server .. but required to throttle the network to avoid full use of network
 +
** IPFS vs WebTorrent:
 +
*** WebTorrent is javascript implementation and require a tab open
 +
*** IPFS: Distributed hash table of encrypted chunk of data

Latest revision as of 14:36, 23 May 2017

Home < Developer Meetings < 20170523


Update

  • NA

To Discuss

  • Use of GDCM vs DCMTK in Slicer
  • Definitive transition from mailing list to discourse

Conclusion

  • slicer.org on GitHub
    • It makes sense to store website on GitHub and have it served from Digital Ocean (or legacy server)
    • Waiting final transition to server is done, cronjob
  • ITK: What ITK should create ?
    • Thin wrapper around DCMTK
    • Default behavior: if context is not known. Secondary capture seems to be the only valid approach.
    • Moving toward consolidating tool around DCMTK
    • Anonymization: David Clunie has implementation in Java. See http://www.dclunie.com/pixelmed/software/webstart/DicomCleanerUsage.html
    • vtk-dicom from David Gobbi is nice. Issue is that it introduces dependency on VTK for tools that wouldn't require it (e.g CLI) + VTK doesn't handle orientation (by default) in its core image data structure
    • DCMTK: currently support only to create secondary capture
    • DCMQI: support to create seg object
  • Public repo with collection of DICOM
    • During last project week, discussed use of IPFS to store medical test data (DICOM, ...)
    • Stored at MIT, Bremen, Standford, ...
    • You can contribute without being the central server
    • It is easy to start an IPFS server .. but required to throttle the network to avoid full use of network
    • IPFS vs WebTorrent:
      • WebTorrent is javascript implementation and require a tab open
      • IPFS: Distributed hash table of encrypted chunk of data