Difference between revisions of "Documentation/Nightly/Modules/DiceComputation"

From Slicer Wiki
Jump to: navigation, search
Line 28: Line 28:
 
<!-- ---------------------------- -->
 
<!-- ---------------------------- -->
 
{{documentation/{{documentation/version}}/module-section|Use Cases}}
 
{{documentation/{{documentation/version}}/module-section|Use Cases}}
{{documentation/modulename}} could be useful to quantitatively compare several registered segmented volumes. It could be used to compare several algorithms with ground truth, or automated segmentation vs manual, etc...
+
{{documentation/modulename}} compute DSC to quantitatively compare several registered segmented volumes. It could be used to compare:
 +
* Several algorithms with ground truth
 +
* Automated segmentation vs manual
 +
* etc...
  
 
<!-- ---------------------------- -->
 
<!-- ---------------------------- -->

Revision as of 14:43, 3 June 2013

Home < Documentation < Nightly < Modules < DiceComputation

Introduction and Acknowledgements

This work is supported by NA-MIC, NCIGT, and the Slicer Community.
Author: Laurent Chauvin, Nobuhiko Hata
Contact: Laurent Chauvin <email> lchauvin@bwh.harvard.edu</email>

NA-MIC  
NCIGT  
SPL  

This project is supported by National Institute of Health (5P01CA067165, 5R01CA124377, 5R01CA138586, 2R44DE019322, 7R01CA124377, 5R42CA137886, 8P41EB015898).

Module Description

This module is capable of calculating the Dice Similarity Coefficient (DSC) between several label map images. If the label map images represent registered segmented structures then the DSC will provide a volumetric alignment between the images. The DSC gives measures of the volumetric overlap between two of the segmented structures, and indicates twice the number of voxels which are shared by or are common to both structures divided by the total number of non-zero voxels in both structures. The DSC can range from zero to one, where zero is no alignment between images and one is perfect alignment. DiceComputation is a Slicer4 module computing DSC for several label map images, and presenting results in a NxN matrix (where N is number of label map images) with color coding for convenient reading.

Use Cases

DiceComputation compute DSC to quantitatively compare several registered segmented volumes. It could be used to compare:

  • Several algorithms with ground truth
  • Automated segmentation vs manual
  • etc...

Tutorials

N/A

Panels and their use

DiceComputation GUI
  1. Parameters
    • Number of label map images to use for computation
  2. Label Maps
    • Select labels map images (only label maps could be selected) to use for computation or None
    • Compute Dice coefficient: Press button to start computation
  3. Results
    • Once computation is over, results will appear in an NxN matrix with color coding
      • Red stripes indicate no label map have been selected (None)
      • Green stripes indicate this is Dice's coefficient of a label map with itself (always 1.0, so not displayed)
      • Values represent Dice's similarity coefficient. The bigger the value is, the greener the background is, in order to quickly find higher matches.


Similar Modules

N/A

References

Information for Developers