Boundary learning by optimization with topological constraints

• Main Authors: Helmstaedter, Moritz N. (Author), Briggman, Kevin L. (Author), Denk, Winfried (Author), Bowden, Jared B. (Author), Mendenhall, John M. (Author), Abraham, Wickliffe C. (Author), Harris, Kristen M. (Author), Kasthuri, Narayanan (Author), Hayworth, Kenneth J. (Author), Schalek, Richard (Author), Tapia, Juan Carlos (Author), Lichtman, Jeff W. (Author), Jain, Viren (Contributor), Bollmann, Benjamin (Contributor), Richardson, Mark A. (Contributor), Berger, Daniel R. (Contributor), Seung, H. Sebastian (Contributor)
• Other Authors: Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences (Contributor)
• Format: Article
• Language: English
• Published: Institute of Electrical and Electronics Engineers (IEEE), 2012-06-27T16:45:42Z.
• Subjects: Article
• Online Access: Get fulltext

 Recent studies have shown that machine learning can improve the accuracy of detecting object boundaries in images. In the standard approach, a boundary detector is trained by minimizing its pixel-level disagreement with human boundary tracings. This naive metric is problematic because it is overly sensitive to boundary locations. This problem is solved by metrics provided with the Berkeley Segmentation Dataset, but these can be insensitive to topological differences, such as gaps in boundaries. Furthermore, the Berkeley metrics have not been useful as cost functions for supervised learning. Using concepts from digital topology, we propose a new metric called the warping error that tolerates disagreements over boundary location, penalizes topological disagreements, and can be used directly as a cost function for learning boundary detection, in a method that we call Boundary Learning by Optimization with Topological Constraints (BLOTC). We trained boundary detectors on electron microscopic images of neurons, using both BLOTC and standard training. BLOTC produced substantially better performance on a 1.2 million pixel test set, as measured by both the warping error and the Rand index evaluated on segmentations generated from the boundary labelings. We also find our approach yields significantly better segmentation performance than either gPb-OWT-UCM or multiscale normalized cut, as well as Boosted Edge Learning trained directly on our data.