Workshop 5, The neuroinformatics of neural connectivity

Albert Cardona

Title: The real neurons: imaging, reconstructing and identifying neurons from serial sections

University of Zurich/ETH Zurich, Zurich, Switzerland

Abstract: Function follows form; precise neuroanatomical data will let us ask the right questions about the information processing capabilities of neural systems. The process of neuronal reconstruction consists of the composition of accurate three-dimensional models of neurons, with detailed dendritic and axonal arbors detailing the circuit wire diagram.

The only available technology for the nanometer scale reconstruction of neuronal tissue is serial section electron microscopy. While image acquisition speed at the electron microscope has advanced by 3 orders of magnitude over the last decade, the state of the art is still limited to relatively small volumes under milimeter size. The imaging technology is the prime limiting factor, and constrains practical neuronal reconstruction to very small volumes. We thus encounter the present´choice: either to concentrate our work on small volumes of large brains, such as thus of vertebrates, or shift our analysis to complete systems of the appropriate size such as the model organism Drosophila melanogaster. The second major challenge is the image segmentation of all sectioned axons and dendrites, and the recognition of synapses. The third challenge is the correlation of the state of the art micrometer-scale neuroanatomical knowledge on insect brains (mostly sparse imaging at the laser-scanning microscope) with the nanometer-scale delivered by electron microscopy. For the purpose of studying the complete brain of the Drosophila larva, myself and others in the neuronal reconstruction community have engaged in the technological development effort towards the automatization of solutions to these challenges, including novel image registration strategies, image recognition algorithms for segmentation, and structural comparison algorithms for the purpose of neuronal recognition and wiring diagram analysis.

Bio sketch: Albert Cardona is a junior group leader at the Institute of Neuroinformatics of he University of Zurich / ETH Zurich. A molecular biologist and neuroanatomist by training, A.C developed a very strong standing in software programming. A.C started and leads two multinational open source software projects for image processing: Fiji and TrakEM2; both focused on image processing, neuron reconstruction and neuroanatomical data analysis.

A.C research targets the understanding of information integration and flow at sensory processing centers of the larva of the fruit fly Drosophila. The approach includes serial section electron microscopy and the 3D reconstruction of neuronal microarchitecture, i.e. the fine structure of neuronal arborizations and synaptic connections, coupled with algorithmic identification of neuronal components by comparison with light-microscopy image data of genetically labeled samples.