MICrONS: Machine Intelligence from Cortical Networks

A ground truth dataset for 3D neuron segmentation from electron microscopy (EM) images of mouse visual cortex, created for IARPA's Machine Intelligence from Cortical Networks (MICrONS) program. For more information, please visit our MICrONS explorer website.

Citation

Functional connectomics spanning multiple areas of mouse visual cortex
MICrONs Consortium et al.
bioRxiv 2021.07.28.454025; doi: https://doi.org/10.1101/2021.07.28.454025

Data Format Specification

Volumes and annotations are stored in a single HDF5 file with the following datasets:

• volumes: are stored in row-major format, i.e., with dimensions (z, y, x). Every volume has an attribute resolution that specifies the voxel size in nm.
  • image
  • segmentation
  • mitochondria (optional)
  • synapse (optional): annotation for postsynaptic densities (PSDs)
  • fold (optional): annotation for folds (a kind of image defect) in images
  • mask: valid/invalid voxels are represented as 1s/0s respectively.
• annotations: are incomplete, meaning that each volume may or may not contain proper annotations.
  • object_ids (optional): small unidentified objects or intracellular organelles that are oversegmented, which may require special handling.
  • myelin_ids (optional)
  • blood_vessel_ids (optional)
  • soma_ids (optional)
  • nucleus_ids (optional)