Difference between revisions of "2011 Winter Project Week:MIDAS for data hosting"

From NAMIC Wiki
Jump to: navigation, search
 
(4 intermediate revisions by the same user not shown)
Line 12: Line 12:
  
 
<h3>Objective</h3>
 
<h3>Objective</h3>
Make use of the [http://www.kitware.com/midaswiki/index.php/Main_Page MIDAS] data archiving system to host data related to tests such as baseline and comparison images.
+
Make use of the [http://www.kitware.com/midaswiki/index.php/Main_Page MIDAS] data archiving system to host data related to tests such as baseline and comparison images.
[http://public.kitware.com/Wiki/TubeTK TubeTK] is a new open-source toolkit that hosts algorithms for applications involving images of tubes.
 
  
Two driving applications:
+
Motivation:
* Surgical guidance: registering pre-operative vascular models with intra-operative images (e.g., ultrasound)
+
* Not all users will want to build and test every module of Slicer, so they shouldn't have to download the testing data for every part of Slicer
* Characterizing vascular patters: using graph theory to distinguish clinical populations based on vascular patterns (e.g., benign -vs- malignant tumors via tortuosity)
+
* Encourage the optional use of testing with "large" datasets that aren't downloaded unless the "large-data" tests are configured to run
 
+
* Don't download any duplicated test data more than once
History
 
* June 2001, UNC released the patent on vessel extraction method from [Aylward, Bullitt 1996...]
 
* TubeTK released under Apache 2.0 license: includes rights to patents
 
  
 
</div>
 
</div>
Line 28: Line 24:
  
 
<h3>Approach, Plan</h3>
 
<h3>Approach, Plan</h3>
* Python module in Slicer 4 for centerline and radius estimation of vasculature in brain MRA
+
* MIDAS+CTest: create a thin wrapper for "add_test" calls that will allow for adding a test that references remote data on a MIDAS server in a human-readable way
** Workflow: brain envelop segmentation, seeding, extraction
+
* The underlying mechanism will address data by its checksum to ensure integrity and no duplication of downloads, but test creators and users shouldn't have to deal with checksums
* Integration with VMTK
+
* Create an "Easy Upload" page for the Slicer community to allow test data to be uploaded and ready to test against with as few clicks as possible
 +
* MIDASDesktop: A GUI application for moving data between a client machine and a MIDAS server
  
 
</div>
 
</div>
Line 37: Line 34:
  
 
<h3>Progress</h3>
 
<h3>Progress</h3>
 +
* MIDAS+CTest macro has been created and is currently in use successfully with the BRAINSTools test data.
 +
* Easy Upload page has a working prototype and requires further improvements and customization for Slicer
 +
* MIDASDesktop: svn repo is here [https://www.kitware.com/svn/KWPublic/trunk/MIDAS/MIDAScpp https://www.kitware.com/svn/KWPublic/trunk/MIDAS/MIDAScpp]
  
 
</div>
 
</div>
Line 45: Line 45:
 
==Delivery Mechanism==
 
==Delivery Mechanism==
  
* Test data will be hosted publicly (with any necessary access control) on [http://midas.kitware.com]
+
* Test data will be hosted publicly (with any necessary access control) on [http://midas.kitware.com midas.kitware.com]
* That site will also host the Easy Upload page.
+
* Test data in the Slicer4 repository will be replaced by a reference
* MIDASDesktop lives in the svn repository here: [https://www.kitware.com/svn/KWPublic/trunk/MIDAS/MIDAScpp]
+
* That site will also host the Slicer Easy Upload page.
  
 
</div>
 
</div>

Latest revision as of 15:08, 10 January 2011

Home < 2011 Winter Project Week:MIDAS for data hosting

Key Investigators

  • Kitware: Zach Mullen
  • UIowa: Hans Johnson

Objective

Make use of the MIDAS data archiving system to host data related to tests such as baseline and comparison images.

Motivation:

  • Not all users will want to build and test every module of Slicer, so they shouldn't have to download the testing data for every part of Slicer
  • Encourage the optional use of testing with "large" datasets that aren't downloaded unless the "large-data" tests are configured to run
  • Don't download any duplicated test data more than once

Approach, Plan

  • MIDAS+CTest: create a thin wrapper for "add_test" calls that will allow for adding a test that references remote data on a MIDAS server in a human-readable way
  • The underlying mechanism will address data by its checksum to ensure integrity and no duplication of downloads, but test creators and users shouldn't have to deal with checksums
  • Create an "Easy Upload" page for the Slicer community to allow test data to be uploaded and ready to test against with as few clicks as possible
  • MIDASDesktop: A GUI application for moving data between a client machine and a MIDAS server

Progress

  • MIDAS+CTest macro has been created and is currently in use successfully with the BRAINSTools test data.
  • Easy Upload page has a working prototype and requires further improvements and customization for Slicer
  • MIDASDesktop: svn repo is here https://www.kitware.com/svn/KWPublic/trunk/MIDAS/MIDAScpp

Delivery Mechanism

  • Test data will be hosted publicly (with any necessary access control) on midas.kitware.com
  • Test data in the Slicer4 repository will be replaced by a reference
  • That site will also host the Slicer Easy Upload page.