Introducing Dice, Jaccard, and Other Label Overlap Measures To ITK
Tustison N., Gee J.
PICSL, University of Pennsylvania
logo

Please use this identifier to cite or link to this publication: http://hdl.handle.net/10380/3141
Although the KappaStatisticImageToImageMetric can be used to obtain the Dice metric (or mean overlap), there are other related measurements that are useful for evaluating results derived from various image analysis tasks. These measures include the target overlap, union overlap (or Jaccard coefficient), and false positive/negative errors. There are also related measures for multilabeled images. The class included with this submission, the itkLabelOverlapMeasuresImageFilter, is meant to provide an easy mechanism for calculating such measures.
Code
plus Automatic Testing Results by Insight-Journal Dashboard on Tue Dec 1 20:56:19 2009 for revision #1
starstarstarstarstar expertise: 5 sensitivity: 5

Reviews
minus Much needed contribution by Dan Mueller on 2010-07-12 12:15:46 for revision #2
starstarstarstarstar expertise: 3 sensitivity: 5
yellow
Summary:

The authors present a filter which computes a number of label overlap measures, including the Dice coefficient. The filter operates on one or more labels.

Evidence:

Full source code and unit tests are provided.

Open Science:

The full source code, unit tests, and input data are provided.

Reproducibility:

I downloaded the code, compiled it, and ran it. I was able to reproduce the results reported in the paper.

Use of Open Source Software:

The filter makes use of the Insight Toolkit.

Open source Contributions:

The article has a clear example how to use the filter.

Code Quality :

The code is of high quality and meets the ITK coding guidelines.

Interest:

The label overlap measures will be of great interest to the Insight community. The contribution makes it simple to automatically compare binary segmentation results, which is helpful to tune parameters and/or compare different algorithms.

Free comment :

This contribution should be merged in to the toolkit ASAP.

Add a new review

Statistics
backyellow
Global rating: starstarstarstarstar
Review rating: starstarstarstarstar [review]
Code rating: starstarstarstarstar
Views: 2440
Downloads: 996

Send a message to the author

Information
backyellow
Paper Id: 707
Categories: Registration, Segmentation
Keywords: Dice, Jaccard, overlap,
Toolkit: ITK
Revision:
Status: Open for public review
View license
Loading license...

Data
backyellow
Full download: .zip
Paper: view, .pdf
Source code : Download

Share
backyellow
Facebook Digg delicious StumbleUpon dzone Furl Technorati Reddit


main_flat
main_bottom
Powered by Midas