TY - JOUR
T1 - Crowdsourcing the creation of image segmentation algorithms for connectomics
AU - Carreras, Ignacio Arganda
AU - Turaga, Srinivas C.
AU - Berger, Daniel R.
AU - San, Dan Cire
AU - Giusti, Alessandro
AU - Gambardella, Luca M.
AU - Schmidhuber, Jürgen
AU - Laptev, Dmitry
AU - Dwivedi, Sarvesh
AU - Buhmann, Joachim M.
AU - Liu, Ting
AU - Seyedhosseini, Mojtaba
AU - Tasdizen, Tolga
AU - Kamentsky, Lee
AU - Burget, Radim
AU - Uher, Vaclav
AU - Tan, Xiao
AU - Sun, Changming
AU - Pham, Tuan D.
AU - Bas, Erhan
AU - Uzunbas, Mustafa G.
AU - Cardona, Albert
AU - Schindelin, Johannes
AU - Seung, H. Sebastian
N1 - Generated from Scopus record by KAUST IRTS on 2022-09-14
PY - 2015/9/5
Y1 - 2015/9/5
N2 - To stimulate progress in automating the reconstruction of neural circuits, we organized the first international challenge on 2D segmentation of electron microscopic(EM) images of the brain. Participants submitted boundary map spredicted for a test set of images, and were scored based on their agreement with a on sensus of human expert annotations. The winning team had no prior experience with EM images, and employed a convolutional network. This “deeplearning” approach has since become accepted as a standard for segmentation of EM images. The challenge has continued to accept submissions, and the best so far has resulted from co-operation between two teams. The challenge has probably saturated, as algorithms cannot progress beyond limits set by ambiguities inherent in 2D scoring and the size of the test data set. Retrospective evaluation of the challenges coring system reveals that it was not sufficiently robust to variations in the widths of neurite borders. We propose a solution to this problem, which should be useful for a future 3D segmentation challenge.
AB - To stimulate progress in automating the reconstruction of neural circuits, we organized the first international challenge on 2D segmentation of electron microscopic(EM) images of the brain. Participants submitted boundary map spredicted for a test set of images, and were scored based on their agreement with a on sensus of human expert annotations. The winning team had no prior experience with EM images, and employed a convolutional network. This “deeplearning” approach has since become accepted as a standard for segmentation of EM images. The challenge has continued to accept submissions, and the best so far has resulted from co-operation between two teams. The challenge has probably saturated, as algorithms cannot progress beyond limits set by ambiguities inherent in 2D scoring and the size of the test data set. Retrospective evaluation of the challenges coring system reveals that it was not sufficiently robust to variations in the widths of neurite borders. We propose a solution to this problem, which should be useful for a future 3D segmentation challenge.
UR - http://journal.frontiersin.org/Article/10.3389/fnana.2015.00142/abstract
UR - http://www.scopus.com/inward/record.url?scp=84948763339&partnerID=8YFLogxK
U2 - 10.3389/fnana.2015.00142
DO - 10.3389/fnana.2015.00142
M3 - Article
SN - 1662-5129
VL - 9
SP - 1
EP - 13
JO - Frontiers in Neuroanatomy
JF - Frontiers in Neuroanatomy
IS - November
ER -