Project: Connectomics with Electron Microscopy


Connectomics with Electron Microscopy


CRCNS EB 005832


Detection of neuron membranes in electron microscopy images

Studying nervous systems via the connectome, the map of connectivities of all neurons in that system, is a challenging problem in neuroscience. In working towards a solution, neurobiologists have acquired large-electron microscopy datasets. However, the shear volume of these datasets renders manual analysis nearly impossible. Hence, automated image analysis methods are required for reconstructing the connectome from these very large image collections. Segmentation of neurons in these images, an essential step of the reconstruction pipeline, is also difficult because of noise, anisotropic shapes and brightness, and the presence of confounding structures. The method we’ve developed uses a series of artificial neural networks (ANNs) in a framework combined with a feature vector that is composed of image intensities sampled over a stencil neighborhood. Several ANNs are applied, allowing each ANN to use the classification context provided by the previous network to improve detection accuracy. We’ve developed the method of serial ANNs to show that the learned context improves detection over traditional ANNs. We also present specific advantages over previous membrane-detection methods. The results are a significant step towards an automated system for the reconstruction of the connectome.



  • E. Jurrus, A.R.C. Paiva, S. Watanabe, J.R. Anderson, B.W. Jones, R.T. Whitaker, E.M. Jorgensen, R.E. Marc, T. Tasdizen. “Detection of Neuron Membranes in Electron Microscopy Images Using a Serial Neural Network Architecture,” In Medical Image Analysis, Vol. 14, No. 6, pp. 770--783. 2010. PubMed ID: 20598935


Automatic mosaicking and volume assembly of Microscopy Images

We present a computationally efficient, robust, and fully automatic method for large-scale electron and confocal microscopy image registration. The proposed method is able to construct large image mosaics from thousands of smaller, overlapping tiles with unknown or uncertain positions, and then align sections from a serial section capture into a common coordinate system. The method also accounts for nonlinear deformations, both in constructing sections and in aligning sections to each other. The publicly available software tools include the algorithms and a Graphical User Interface for easy access to the algorithms.


See our Publications for the latest research.

Quick Links


Related Research Themes