Crowdsourcing the creation of image segmentation algorithms for connectomics

By Ignacio Arganda-Carreras, Srinivas C. Turaga, D. Berger, Dan Claudiu Ciresan, Alessandro Giusti, Luca Maria Gambardella, J. Urgen Schmidhuber, Dmitry Laptev, Sarvesh Dwivedi, Joachim M. Buhmann and 14 more. Frontiers in Neuroanatomy, 2015

371 citations recorded by the source corpus.

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 maps predicted for a test set of images, and were scored based on their agreement with a consensus of human expert annotations. The winning team had no prior experience with EM images, and employed a convolutional network. This “deep learning” 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 cooperation 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 dataset. Retrospective evaluation of the challenge scoring 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.

Subfields: Biophysics, Materials Chemistry, Structural Biology

View Paper: https://www.semanticscholar.org/paper/e07e61de05d30a348cb7b4cb38aede541f80941c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biophysics

In Materials Chemistry

In Structural Biology

Citations in the corpus

Democratising deep learning for microscopy with ZeroCostDL4Mic (2021, methodology): “For 2D segmentation, this was a neuronal membrane segmentation dataset from the ISBI challenge 2012 78,79 and for 3D segmentation from the mitochondrial segmentation dataset available from EPFL (https://www.ep ﬂ .ch/labs/cvlab/data/data-em/).”
Learning Steerable Filters for Rotation Equivariant CNNs (2017, methodology): “(ii) A processing pipeline based on the proposed SFCNN layers ranks among the top three entries in the ISBI 2012 electron microscopy segmentation challenge [5].”
Suggestive Annotation: A Deep Active Learning Framework for Biomedical Image Segmentation (2017, background): “Recent advances in deep learning [2,3,12,15,16] have achieved promising results on many biomedical image segmentation benchmarks [1,14].”
CubeNet: Equivariance to 3D Rotation and Translation (2018, methodology): “Scoring is performed using the metrics Vrand and Vinfo described in [1].”
Superhuman Accuracy on the SNEMI3D Connectomics Challenge (2017, methodology): “The SNEMI3D leaderboard measures segmentation quality based on the adapted Rand F-score [32, 33].”
CE-Net: Context Encoder Network for 2D Medical Image Segmentation (2019, background): “The specific computation process and more details of these two algorithms could be found in [76].”
CLoDSA: a tool for augmentation in classification, localization, detection, semantic segmentation and instance segmentation tasks (2019, methodology): “In particular, we tackle the automatic segmentation of neural structures using the dataset from the ISBI challenge [44].”
Rotation Equivariant Vector Field Networks (2016, methodology): “Results: A detailed explanation on the evaluation metrics used in the challenge can be found in the ISBI 2012 challenge website 1 and in [1].”
MultiResUNet : Rethinking the U-Net Architecture for Multimodal Biomedical Image Segmentation (2019, methodology): “To observe the eﬀectiveness of the architecture with electron microscopy images, we used the dataset of the ISBI-2012 : 2D EM segmentation challenge [40, 41].”
Evolutionary Compression of Deep Neural Networks for Biomedical Image Segmentation (2020, neutral): “scores can be found in [64].”
Self-Supervised Representation Learning by Rotation Feature Decoupling (2019, background): “They are not favored in a rotation agnostic image dataset like plankton [8] and ISBI 2012 electron microscopy segmentation challenge [3].”
Deep convolutional neural networks for brain image analysis on magnetic resonance imaging: a review (2017, methodology): “The network was successfully tested on the 2D electron microscopy [141] and ISBI 2015 cell tracking6 challenges, achieving the first position in both of them.”
UNCERTAINTY MODELING AND INTERPRETABILITY IN CONVOLUTIONAL NEURAL NETWORKS FOR POLYP SEGMENTATION (2018, background): “Database Training Validation Test Resolution CVC-Colon [5] [1-76], [98-148], [221-273].”
DeepEM3D: approaching human‐level performance on 3D anisotropic EM image segmentation (2017, methodology): “, 2016), we developed deep learning methods for segmenting 2D EM images (Arganda-Carreras et al., 2015).”
The Importance of Skip Connections in Biomedical Image Segmentation (2016, methodology): “In this section, we test the model on electron microscopy (EM) data [2] (Section 3.”
U-Net and Its Variants for Medical Image Segmentation: A Review of Theory and Applications (2020, neutral): “segmentation challenge [271].”
Topology-Preserving Deep Image Segmentation (2019, methodology): “3 Experiments We evaluate our method on six natural and biomedical datasets: CREMI7, ISBI12 [4], ISBI13 [3], CrackTree [45], Road [26] and DRIVE [36].”