Convolutional nets for reconstructing neural circuits from brain images acquired by serial section electron microscopy
arXiv:1904.12966
Abstract
Neural circuits can be reconstructed from brain images acquired by serial section electron microscopy. Image analysis has been performed by manual labor for half a century, and efforts at automation date back almost as far. Convolutional nets were first applied to neuronal boundary detection a dozen years ago, and have now achieved impressive accuracy on clean images. Robust handling of image defects is a major outstanding challenge. Convolutional nets are also being employed for other tasks in neural circuit reconstruction: finding synapses and identifying synaptic partners, extending or pruning neuronal reconstructions, and aligning serial section images to create a 3D image stack. Computational systems are being engineered to handle petavoxel images of cubic millimeter brain volumes.
References in corpus (14)
- Semantic Instance Segmentation with a Discriminative Loss Function
- Superhuman Accuracy on the SNEMI3D Connectomics Challenge
- Semantic Instance Segmentation via Deep Metric Learning
- ssEMnet: Serial-section Electron Microscopy Image Registration using a Spatial Transformer Network with Learned Features
- Image Segmentation by Size-Dependent Single Linkage Clustering of a Watershed Basin Graph
- A Multi-Pass Approach to Large-Scale Connectomics
- RhoanaNet Pipeline: Dense Automatic Neural Annotation
- Morphological Error Detection in 3D Segmentations
- Identifying Synapses Using Deep and Wide Multiscale Recursive Networks
- Siamese Encoding and Alignment by Multiscale Learning with Self-Supervision
- Chunkflow: Distributed Hybrid Cloud Processing of Large 3D Images by Convolutional Nets
- Deep Learning Improves Template Matching by Normalized Cross Correlation
- Toward Streaming Synapse Detection with Compositional ConvNets
- Adversarial Image Alignment and Interpolation