FlyEM Connectome Project EM Neural Circuits Source Data

At Janelia, the FlyEM Project is developing technology that will enable us to create a map of every neuron and synapse in the Drosophila nervous system, using novel approaches to electron microscopy (EM) as the foundation. Researchers can use the data offered here to map neural circuitry. 

Fly medulla small ground truth volumes

This repository contains several small subvolumes of EM data from the Drosophila optic lobe medulla. These are regions from the same dataset as medulla7column in the DICED gs://flyem-public-connectome, but the ground truth is more accurate. These examples are documented in more detail in neuroproof_examples.

The ground truth is not pixel-perfect but should be suitable for supervised training.

Each subvolume is listed below. For a given volume X, there is a 3D grayscale volume X-grayscale and a ground truth volume X-ground truth.

training2: 250x250x250 subvolume (corresponds to training_sample2 in neuroproof_examples)

val: 500x500x500 subvolume (corresponds to validation_sample in neuroproof_examples)

Seven medulla column connectome

This repository contains the EM image source data, segmentation, and proofread result for seven medulla columns in the Drosophila optic lobe. Researchers can read information about this connectome and some of the interesting biological insights here.

The following describes the different 3D image arrays in this package and the various support files. This dataset should be conducive for studying 3D neuronal morphology, connectomes, and as ground truth for validating EM segmentation and synapse prediction algorithms. This package specifies a smaller ROI within this dataset where the reconstruction accuracy is high. The completeness is not 100% in this region, so it might be better for validation instead of supervised classifier training.

The results correspond to the internal DVID version of #3b548. For convenience, that repository was flattened and copied into this one.

Central complex small ground truth volumes

This repository contains several small subvolumes of EM data from the Drosophila central complex. Each subvolume of raw grayscale data is accompanied by segmentation labels data.

The ground truth is not pixel-perfect but should be suitable for supervised training.

Each subvolume is described below. For a given volume X, there is a 3D grayscale volume X-grayscale and a ground truth volume X-ground truth. Some of the segmentation/body IDs have been identified as glia.

pb1: ~500x500x500 subvolume in the fly protocerebral bridge. Body id 16232 represents glia.

pb2: ~500x500x500 subvolume in the fly protocerebral bridge. Body id 16280 represents glia.