Dataset · 2013
Helmstaedter et al. — mouse retinal connectome
A dense reconstruction of the inner plexiform layer, and the first demonstration that distributed human annotation could scale EM reconstruction.
| Species | Mouse |
|---|---|
| Region | Retina (inner plexiform layer) |
| Modality | SBEM with crowd-sourced annotation |
| Access | Open |
| Status | Released |
Helmstaedter and colleagues densely reconstructed a volume of mouse inner plexiform layer using serial block-face EM, with a large team of trained annotators. It is the origin of the modern proofreading-as-labour problem: the paper is as much about how the annotation was organised as about the circuit.
What it is good for. The historical argument for automation. Read it alongside the segmentation unit to see what the field was doing before flood-filling networks.
What it does not support. Direct comparison of its error rates to modern pipelines; the annotation protocol and the quality metrics are not the ones in use today.
Where these numbers come from. Helmstaedter et al. (2013), Nature 10.1038/nature12346; BossDB. DOI 10.1038/nature12346.