Our brains are composed of billions of nerve cells and each of those cells can be connected to thousands of other individual cells, making for an incredibly complex network. While three-dimensional electron microscopy is used to image cells and their connections, typically manual observation and mapping is used to create the connectivity diagrams of small areas of the brain. In order to build the “connectome” of the brain, computers will have to do the job of transforming images obtained through microscopy into a usable 3D map that scientists can utilize for all kinds of applications.
Researchers at Google AI and the Max Planck Institute of Neurobiology have now developed a way of turning microscopy image stacks into maps of individuals neurons, including their connections. The technology utilizes artificial neural networks to complete the feat, and the results closely track with what manual mapping by trained individuals would result in.
Study in Current Opinion in Neurobiology: Progress and remaining challenges in high-throughput volume electron microscopy…