Sign In

Communications of the ACM

ACM TechNews

Carnegie Mellon Develops New Method For Analyzing Synaptic Density

View as: Print Mobile App Share:
Electron micrograph of stained somatosensory cortex synapses that were identified using a machine-learning algorithm.

Carnegie Mellon University researchers have developed a new approach to broadly survey learning-related changes in synapse properties.

Credit: Saket Navlakha, Alison L. Barth

Carnegie Mellon University (CMU) researchers say they are using machine learning to survey learning-related changes in synapse properties.

The researchers analyzed thousands of images from the cerebral cortex, which enabled them to identify synapses from an entire cortical region, revealing new information about how synaptic properties change during development and learning.

Traditional electron microscopy, when used to study a large section of the brain, would result in terabytes of data, but the researchers say their technique simplifies this problem by combining a specialized staining process with machine learning. The method produces lower-resolution information from a larger region of the brain, instead of perfect information from a tiny part of the brain. The technique uses a special chemical preparation, which deeply stains the synapses in a sample of brain tissue, and then the researchers use machine-learning algorithms to identify and compare synapse properties across a column of the cerebral cortex.

They tested the technique by examining how synapses across a complex circuit would change with altered somatosensory input. The researchers analyzed nearly 25,000 images and 40,000 synapses and found the method could be used to determine increases in synapse density and size during development and learning. In addition, they found synapse properties changed in a coordinated way across the entire region of the neocortex examined.

From Carnegie Mellon News (PA)
View Full Article


Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account