Home > SCI Research: Collaborative Research in Computational Neuroscience (CRCNS)


Back to Research

NIH/NSF Collaborative Research in Computational Neuroscience (CRCNS)

Video Clip [38Mb] Overview of the CRCNS Project
The answers to many important biological questions depend on a better understanding of cellular ultrastructure, which forms the interface between biochemistry and anatomy. Biological imaging (i.e. microscopy image acquisition) is progressing at an extraordinary pace, and biologists are finding that the bottleneck in producing the next generation of scientific achievements is not the acquisition of images, but the analysis of those images. Serial section microscopy is one of the most compelling examples of this gap between image acquisition and analysis. Detailed, data-driven descriptions of microscopic structures are especially important in neurobiology. While neural modeling is critical to our understanding of the central nervous system, state-of-the-art models are relatively unconstrained by anatomical data. Very little is known about the physical organization and connectivities of neurons.

Motivated by the need for better biological image analysis, we have launched the NIH/NSF Collaborative Research in Computational Neuroscience (CRCNS) project bringing experts in Scientific Computing at the SCI Institute together with researchers in ophthalmology from the Robert Marc Laboratory at the University of Utah's Moran Eye Center and researchers in neurobiology from the Chi-Bin Chien Laboratory at the University of Utah's Department of Neurobiology & Anatomy. Funded by the National Institutes of Health (NIH) and the National Science Foundation (NSF), the CRCNS project aims to develop new methods for assembling many 2D images into larger mosaics and many layers into 3D volumes and then automatically isolating structures from within the volume.

3D rendering of 4 axons tracked through a volume of ~300 slices.

Tracking a single axion through sequential layers of a specimen of rabbit retina.

Serial section microscopy takes a series of slices through a specimen.
Our initial case of study is an attempt to reconstruct the wiring diagram of neurons in the rabbit retina and the zebra fish optic tract from serial section microscopy images. Serial section microscopy is a method of taking a series of electron microscopy images or slices of a specimen at regular intervals that can be re-assembled into a 3D volume. This 3D data contains valuable information about structures in the specimen such as dendrites and axons, as well as the anatomical relationships of these structures. The state of the art for tracking structures through slices is manual labor, but with the increasing amount of data generated by modern imaging technologies this method is no longer practical and automated processes is needed. A central aspect of our research the development of automatic segmentation software to allow researcher to automatically isolate structures within the data and find their connectivity. Once the specimen has been segmented, it is possible to build (3D) connectivity maps for neurons in the retinal tissue.

Due to the limited field of view provided by an electron microscope using high magnification, typically a number of image tiles must be assembled together to compose each slice. One of central challenges of this project is to develop the algorithms and software tools needed to automatically register and assemble the multiple images into a mosaic. SCI researchers working on the CRCNS project have developed an automatic image registration algorithm and suite of tools to achieve this task.

Feature based registration uses features in the image itself to align overlapping images.

Many image tiles (indicated by color difference) must be registered and combined to build a mosaic for each slice. Hundreds of such slices are then stacked to build a 3D volume which is used to explore the anatomy of the retina.

Another major challenge is the tracking of features such as axons and their connectivity over relatively long distances within the data. Although the physical scale is very small, a single axon may traverse as many as 1000 cross-sectional slices. Large numbers of axons are tightly packed and they slowly wind in and around other axons along the optic track. Tracking individual axons is a difficult process. SCI researchers have developed methods that successfully track individual axons through hundreds of image slices.

The ability to study and visualize complex cellular structures such as axons and neurons will be a major advancement to neuroscience and the fight against blindness. The CRCNS project promises to significantly advance the cutting edge of tools and techniques needed for research in this area.

Principal Researchers: