# Convolutional nets for reconstructing neural circuits from brain images acquired by serial section electron microscopy

## TL;DR

This paper reviews the application of convolutional neural networks in automating neural circuit reconstruction from serial section electron microscopy images, highlighting progress and ongoing challenges in handling complex brain image data.

## Contribution

It provides an overview of how convolutional nets are used for various tasks in neural circuit reconstruction, emphasizing recent advances and remaining hurdles.

## Key findings

- Convolutional nets have achieved high accuracy in boundary detection on clean images.
- Handling image defects remains a significant challenge.
- Systems are being developed to process petavoxel-scale brain images.

## Abstract

Neural circuits can be reconstructed from brain images acquired by serial section electron microscopy. Image analysis has been performed by manual labor for half a century, and efforts at automation date back almost as far. Convolutional nets were first applied to neuronal boundary detection a dozen years ago, and have now achieved impressive accuracy on clean images. Robust handling of image defects is a major outstanding challenge. Convolutional nets are also being employed for other tasks in neural circuit reconstruction: finding synapses and identifying synaptic partners, extending or pruning neuronal reconstructions, and aligning serial section images to create a 3D image stack. Computational systems are being engineered to handle petavoxel images of cubic millimeter brain volumes.