Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation
Many image segmentation algorithms first generate an affinity graph and then partition it. We present a machine learning approach to computing an affinity graph using a convolutional network (CN) trained using ground truth provided by human experts. The CN affinity graph can be paired with any stand...
| Main Authors: | , , , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | en_US |
| Published: |
MIT Press
2011
|
| Online Access: | http://hdl.handle.net/1721.1/60924 |
| _version_ | 1826214012004597760 |
|---|---|
| author | Turaga, Srinivas C. Murray, Joseph F. Jain, Viren Roth, Fabian Helmstaedter, Moritz N. Briggman, Kevin L. Denk, Winfried Seung, H. Sebastian |
| author2 | Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences |
| author_facet | Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences Turaga, Srinivas C. Murray, Joseph F. Jain, Viren Roth, Fabian Helmstaedter, Moritz N. Briggman, Kevin L. Denk, Winfried Seung, H. Sebastian |
| author_sort | Turaga, Srinivas C. |
| collection | MIT |
| description | Many image segmentation algorithms first generate an affinity graph and then partition it. We present a machine learning approach to computing an affinity graph using a convolutional network (CN) trained using ground truth provided by human experts. The CN affinity graph can be paired with any standard partitioning algorithm and improves segmentation accuracy significantly compared to standard hand-designed affinity functions.
We apply our algorithm to the challenging 3D segmentation problem of reconstructing neuronal processes from volumetric electron microscopy (EM) and show that we are able to learn a good affinity graph directly from the raw EM images. Further, we show that our affinity graph improves the segmentation accuracy of both simple and sophisticated graph partitioning algorithms.
In contrast to previous work, we do not rely on prior knowledge in the form of hand-designed image features or image preprocessing. Thus, we expect our algorithm to generalize effectively to arbitrary image types. |
| first_indexed | 2024-09-23T15:58:23Z |
| format | Article |
| id | mit-1721.1/60924 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T15:58:23Z |
| publishDate | 2011 |
| publisher | MIT Press |
| record_format | dspace |
| spelling | mit-1721.1/609242022-10-02T05:26:57Z Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation Turaga, Srinivas C. Murray, Joseph F. Jain, Viren Roth, Fabian Helmstaedter, Moritz N. Briggman, Kevin L. Denk, Winfried Seung, H. Sebastian Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences Seung, H. Sebastian Turaga, Srinivas C. Murray, Joseph F. Jain, Viren Roth, Fabian Seung, H. Sebastian Many image segmentation algorithms first generate an affinity graph and then partition it. We present a machine learning approach to computing an affinity graph using a convolutional network (CN) trained using ground truth provided by human experts. The CN affinity graph can be paired with any standard partitioning algorithm and improves segmentation accuracy significantly compared to standard hand-designed affinity functions. We apply our algorithm to the challenging 3D segmentation problem of reconstructing neuronal processes from volumetric electron microscopy (EM) and show that we are able to learn a good affinity graph directly from the raw EM images. Further, we show that our affinity graph improves the segmentation accuracy of both simple and sophisticated graph partitioning algorithms. In contrast to previous work, we do not rely on prior knowledge in the form of hand-designed image features or image preprocessing. Thus, we expect our algorithm to generalize effectively to arbitrary image types. 2011-02-11T16:36:49Z 2011-02-11T16:36:49Z 2010-01 Article http://purl.org/eprint/type/JournalArticle 0899-7667 1530-888X http://hdl.handle.net/1721.1/60924 Turaga, Srinivas C. et al. “Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation.” Neural Computation 22.2 (2011): 511-538. © 2009 Massachusetts Institute of Technology en_US http://dx.doi.org/10.1162/neco.2009.10-08-881 Neural Computation Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf MIT Press MIT web domain |
| spellingShingle | Turaga, Srinivas C. Murray, Joseph F. Jain, Viren Roth, Fabian Helmstaedter, Moritz N. Briggman, Kevin L. Denk, Winfried Seung, H. Sebastian Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation |
| title | Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation |
| title_full | Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation |
| title_fullStr | Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation |
| title_full_unstemmed | Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation |
| title_short | Convolutional Networks Can Learn to Generate Affinity Graphs for Image Segmentation |
| title_sort | convolutional networks can learn to generate affinity graphs for image segmentation |
| url | http://hdl.handle.net/1721.1/60924 |
| work_keys_str_mv | AT turagasrinivasc convolutionalnetworkscanlearntogenerateaffinitygraphsforimagesegmentation AT murrayjosephf convolutionalnetworkscanlearntogenerateaffinitygraphsforimagesegmentation AT jainviren convolutionalnetworkscanlearntogenerateaffinitygraphsforimagesegmentation AT rothfabian convolutionalnetworkscanlearntogenerateaffinitygraphsforimagesegmentation AT helmstaedtermoritzn convolutionalnetworkscanlearntogenerateaffinitygraphsforimagesegmentation AT briggmankevinl convolutionalnetworkscanlearntogenerateaffinitygraphsforimagesegmentation AT denkwinfried convolutionalnetworkscanlearntogenerateaffinitygraphsforimagesegmentation AT seunghsebastian convolutionalnetworkscanlearntogenerateaffinitygraphsforimagesegmentation |