Awesome vEM Datasets

A curated list of volume electron microscopy (vEM) datasets, from benchmark cubes to whole-brain and petascale connectomes, with the ground truth, portals, tools and literature needed to use them.

Volume EM in one paragraph. vEM images a resin-embedded, heavy-metal-stained specimen as a stack of thousands of serial electron micrographs, either by repeatedly imaging and shaving the block face (SBEM, FIB-SEM) or by imaging ultrathin serial sections (ssTEM, ATUM-SEM, multibeam SEM, GridTape TEM). The result is a 3D image with nanometre voxels (typically 3–20 nm in-plane and 4–50 nm between slices) in which every membrane, organelle and chemical synapse is visible. That makes vEM the reference method for synaptic-resolution connectomics, the complete wiring diagrams of nervous systems, and a key tool for cell biology. Recognition has followed the scale: Nature listed vEM among its technologies to watch in 2023, and Nature Methods named EM-based connectomics its Method of the Year 2025. Datasets now range from ~10 μm cubes to the whole adult Drosophila central nervous system, whole larval zebrafish brains and cubic millimetres of mouse and human cortex (> 1 PB each). Reconstruction rather than imaging is the bottleneck, which is why the ground truth collected here matters. The primer covers techniques, sample preparation, the reconstruction pipeline and data infrastructure.

Contents

• Landmark Connectomes
• Petascale Volumes
• Benchmarks and Challenges
• From Volume to Benchmark to Connectome
• Accessible Ground Truth
• Pre-training Corpora
• Data Portals and Tools
• Reviews
• Related Lists and Surveys
• Related Code
• Contribution

Page	What is inside
Background	primer on vEM: techniques, sample preparation, from images to connectomes, data scale, programs, figures
Museum of Datasets	every dataset by organism (benchmark cubes, nematodes, insects, other invertebrates, fish, birds, rodents, primates, cell biology) with species, sample, microscopy, size, resolution and links
Reference	the papers behind the datasets, by year, with figures
Research Group	labs and companies that produce vEM datasets and tools
Contributing	how to add or correct an entry

Landmark Connectomes

Complete nervous systems (or brains) and the largest mammalian volumes reconstructed at synaptic resolution; newest first. Details and many more datasets are in the museum.

Year	Organism	Dataset	Microscopy	Scale	Data
2026	Drosophila, adult male CNS (brain + VNC)	Male CNS	FIB-SEM	~166,700 neurons	male-cns
2026	Drosophila, adult female brain + VNC	BANC	ssTEM (GridTape)	brain and nerve cord of one animal	bossdb
2025	Mouse, visual cortex (~1 mm3)	MICrONS	autoTEM	~200,000 cells, 524 M synapses	microns-explorer
2025	Platynereis larva, whole body	Platynereis	ssTEM	>9,000 cells	catmaid
2025	Zebrafish larva, brain + spinal cord	Fish1	–	~187,000 cells, ~30 M synapses (preprint)	fish1
2025	Zebrafish larva, whole brain	Fish-X	ssSEM	~177,000 cells, ~25 M synapses (preprint)	–
2025	Drosophila, adult optic lobe	Optic Lobe	FIB-SEM	~53,000 neurons	neuprint
2024	Drosophila, adult female brain	FAFB / FlyWire	ssTEM (TEMCA2)	139,255 neurons, 54.5 M synapses	codex
2024	Drosophila, adult male VNC	MANC	FIB-SEM	~23,000 neurons	neuprint
2024	Drosophila, adult female VNC	FANC	ssTEM (GridTape)	~14,600 neuronal cell bodies	bossdb
2024	Human temporal cortex (~1 mm3)	H01	ATUM-mSEM	1.4 PB, 104 proofread neurons	h01-release
2023	Drosophila larva, brain	L1EM	ssTEM	3,016 neurons, 548,000 synapses	catmaid
2021	C. elegans, birth to adulthood	Witvliet2020	ssTEM	8 brains	nemanode
2020	Drosophila, adult central brain	Hemi-brain	FIB-SEM	~25,000 neurons	neuprint
2019	C. elegans, both sexes	WormWiring	ssTEM	302 / 385 neurons	wormwiring
2016	Ciona larva, CNS	Ciona Larva	ssTEM	177 neurons	–
1986	C. elegans hermaphrodite	MoW	ssTEM	302 neurons	wormatlas

In progress: the BRAIN CONNECTS mouse hippocampal formation (~10 mm³, multibeam SEM) and Fish Fire&Wire (whole-brain activity + EM of the same larval zebrafish); see Programs and Initiatives.

Petascale Volumes

Year	Name	Size	Link
2024	Si (e.g., Si150L4)	~1.1 PB	https://wklink.org/7122
2021	H01	~1.4 PB	https://h01-release.storage.googleapis.com/data.html
2021	MICrONS (mm3)	~2.0 PB	https://microns-explorer.org/

Rule of thumb: 1 mm³ at 4×4×40 nm is ~1.6×10¹⁵ voxels, i.e. ~1.6 PB of 8-bit raw data.

Benchmarks and Challenges

Year	Name	Task	Data	Venue	Link
2025	CellMap	3D semantic + instance segmentation of >40 organelle classes	FIB-SEM of cells and tissues, 4/8 nm	Janelia	https://cellmapchallenge.janelia.org/
2025	ConnectomeBench	proofreading with (multimodal) LLMs	MICrONS, FlyWire	NeurIPS 2025	https://github.com/jffbrwn2/ConnectomeBench
2024	NISB	neuron instance segmentation	–	Connectomics Conf	https://structuralneurobiologylab.github.io/nisb/#
2023	WASPSYN	domain-adaptive synapse detection	micro-wasp FIB-SEM, 8 nm	ISBI 2023	https://codalab.lisn.upsaclay.fr/competitions/9169
2023	XPRESS X-ray, not EM	myelinated-axon segmentation	mouse white matter, XNH 100 nm	ISBI 2023	https://xpress.grand-challenge.org/
2021	MitoEM	mitochondria instance segmentation	rat + human cortex, mSEM	ISBI 2021	https://mitoem.grand-challenge.org/
2021	AxonEM	axon instance segmentation	mouse + human cortex	MICCAI 2021	https://axonem.grand-challenge.org/
2021	NucMM	nucleus instance segmentation	zebrafish (vEM) + mouse (micro-CT)	MICCAI 2021	https://nucmm.grand-challenge.org/
2016	CREMI	neurons, synaptic clefts, synaptic partners	FAFB ssTEM	MICCAI 2016	https://cremi.org/
2013	SNEMI3D	3D neurite segmentation	Kasthuri15 ATUM-SEM	ISBI 2013	https://snemi3d.grand-challenge.org/
2012	ISBI	2D membrane segmentation	Drosophila VNC ssTEM	ISBI 2012	https://imagej.net/events/isbi-2012-segmentation-challenge

From Volume to Benchmark to Connectome

Many benchmark subsets were cut from larger volumes that were later reconstructed as connectomes. Use this table to find the parent data and to avoid train/test leakage.

Full set	Subset	Connectome	Link
FAFB	CREMI-A/B/C; neurotransmitter labels	FlyWire	https://codex.flywire.ai/
Kasthuri15	SNEMI3D/AC3/AC4	saturated reconstruction (two cylinders)	https://lichtman.rc.fas.harvard.edu/vast/
Hemi-brain	Hemi-brain training set (LSD)	Hemi-brain v1.2.1	https://neuprint.janelia.org/
FIB-25	FIB-25 training set	seven-column medulla	https://neuprint-examples.janelia.org/
J0126	J0126 training set (FFN, LSD “Zebrafinch”)
MICrONS (mm3)	MICrONS training stacks; AxonEM (mouse); ConnectomeBench	MICrONS functional connectome	https://www.microns-explorer.org/cortical-mm3
H01	AxonEM (human); BvEM (human)	104 proofread neurons + automated reconstruction	https://h01-release.storage.googleapis.com/landing.html
Megaphragma whole-brain FIB-SEM	WASPSYN		https://codalab.lisn.upsaclay.fr/competitions/9169
L1EM (larval CNS)		larval Drosophila whole-brain connectome	https://l1em.catmaid.virtualflybrain.org/
Male CNS		male Drosophila CNS connectome (same sample as the optic-lobe dataset)	https://male-cns.janelia.org/
BANC		brain-and-nerve-cord connectome	https://bossdb.org/project/bates_phelps_kim_yang2025
e2198	EyeWire cubes	retinal direction-selectivity circuit + ganglion-cell museum	https://museum.eyewire.org/
e2006	SegEM retina training set	inner plexiform layer dense connectome (950 cells)	https://neuro.rzg.mpg.de/
S1 SBEM (2012-09-28_ex145)	SegEM cortex training set	L4dense connectome	https://l4dense2019.brain.mpg.de/
S1 mSEM (Sievers)		first complete cortical column connectome	https://webknossos.org/publications
MEC SBEM (Schmidt)		axonal synapse sorting in medial entorhinal cortex	https://webknossos.org/publications
Bock2011		V1 functional network	https://bossdb.org/project/bock2011
Lee2016		V1 excitatory network (function + structure)	https://bossdb.org/project/lee2016
dLGN		visual thalamus network	https://bossdb.org/project/morgan2020
Cerebellum (Nguyen)		cerebellar pattern-separation circuit	https://bossdb.org/project/nguyen_thomas2022
Svara2022 (mapzebrain)		whole-brain larval zebrafish reconstruction (~121,000 neurons)	https://mapzebrain.org
Fish1		whole-brain 7-dpf larval zebrafish community connectome	https://fish1-release.storage.googleapis.com/index.html

Accessible Ground Truth

Labeled data for training and evaluating segmentation, detection and proofreading models, grouped by task.

Name for short	Size μm3	Resolution nm3	GT Size μm3	GT Resolution nm3	GT Size voxel	Link	Note
Neurons: dense labels and proofread reconstructions
H01	2,000x3,000x175	4x4x30		8x8x30	104 proofread neurons 183,000,000 synapses	google swc-zip	proofread cells as SWC skeletons + subcellular annotations
Hemi-brain v1.2	~250x250x250	8x8x8		8x8x8	~25,000 neurons ~20,000,000 synapses	janelia neuprint	proofread segmentation + synapses at gs://neuroglancer-janelia-flyem-hemibrain
LSD benchmark volumes					Zebrafinch, Hemi-brain, FIB-25	github aws-registry	training/evaluation volumes bundled with Local Shape Descriptors (Sheridan et al. 2023); Zebrafinch = J0126
e2006 (repository)	114x80x132	16.5x16.5x25		16.5x16.5x25	950 skeletons	mpi-repo	mouse retina (Helmstaedter 2013): EM cubes, skeletons, segmentations, contact matrices, CNN training data
L4dense (repository)	62x95x93	11.24x11.24x28			89 somata 6,979 axons	mpi-webdav	mouse barrel cortex L4 (Motta 2019): HDF5 reconstructions + NML training/validation annotations (webKnossos)
J0126 (training)	96x98x114	9x9x20		9x9x20	33 blocks 12+50 skels	cloudvolume-raw	zebra finch area X, FFN training/evaluation blocks
MICrONS Pinky (training)		4x4x40			3 stacks	bossdb microns zenodo	mouse visual cortex, MICrONS phase 1; zenodo HDF5 with neuron, mitochondria and synapse (PSD) annotations
FIB-25 (training)		8x8x8		8x8x8		ffn-sample neuroproof	drosophila optic medulla, dense GT (also FFN training data)
Wafer (MEC)		8x8x35		8x8x35	1.2 billion voxels labeled (6 regions, 1250x1250x125 each)	huggingface	mouse MEC wafer data of TokenUnify (ICCV 2025); access request required
AxonEM	30x30x30	7x7x40 8x8x30		7x7x40 8x8x30		grand-challenge	subsets of MICrONS and H01
CREMI A/B/C		4x4x40		4x4x40	3x 1250x1250x125	cremi	from FAFB; neuron + synaptic cleft + partner labels
SNEMI3D (AC3/AC4)		6x6x30		6x6x30	AC3: 1024x1024x256 AC4: 1024x1024x100	grand-challenge drive-mirror	subsets of Kasthuri15 (mouse cortex), dense neurites
Harris2015					3 volumes	bossdb	rat hippocampus CA1 neuropil (dendrites, axons, synapses)
ISBI 2012	2x2x1.5	4x4x50	2x2x1.5	4x4x50	30x 512x512	imagej	drosophila VNC, membrane labels
HVC skeletons						github	zebra finch HVC skeleton reconstructions of Kornfeld et al. 2017
STAR (test)						zenodo	generalization test volume of the STAR challenge
Synapses and neurotransmitters
Neurotransmitter labels		4x4x40			180,675 synapses (1,164 neurons, 6 transmitters)	zenodo	FAFB synapses classified as ACh, Glu, GABA, DA, 5-HT or octopamine (Eckstein et al. 2024)
WASPSYN		8x8x8		8x8x8	14x 416x416x416	codalab	micro-wasp FIB-SEM; pre-/postsynaptic points; domain adaptation across 3 brains
FAFB Synapses (synful)		4x4x40			244,000,000 synaptic partners	zenodo	whole-brain synaptic partner predictions of Buhmann et al. 2021
SynapseNet					>117,000 vesicles (337 tomograms)	zenodo	vesicles, active zones, mitochondria in hippocampal synapses; electron tomography rather than serial-section vEM
Organelles, nuclei and blood vessels
CellMap		4x4x4 8x8x8		4x4x4 8x8x8	289 crops from 22 datasets	challenge data github	>40 organelle classes in cells and tissues (FIB-SEM); CellMap Segmentation Challenge 2025
MitoEM 2.0						zenodo github	hard cases (dense, hyperfused, thin mitochondria) across tissues and FIB-SEM / SBEM / ssSEM; official splits
MitoEM	2x 30x30x30	8x8x30	2x 30x30x30	8x8x30	~40,000 instances	grand-challenge	mitochondria, human + rat cortex
CEM-MitoLab / MitoNet benchmarks					21,860 images 135,285 mitochondria	EMPIAR-11037 EMPIAR-10982	2D training set + six 3D benchmark volumes (C. elegans, fly brain, HeLa, muscle, salivary gland, Lucchi++) of Conrad and Narayan 2023
Lucchi (EPFL Hippocampus)		5x5x5		5x5x5	2x 1024x768x165	epfl	mitochondria benchmark (FIB-SEM, hippocampus CA1)
Lucchi++ / Kasthuri++						casser.io	re-annotated mitochondria benchmarks
NucMM						grand-challenge	nuclei; zebrafish whole brain (vEM) + mouse cortex (micro-CT)
BvEM						project bossdb	cortical blood vessels in mouse, macaque and human vEM volumes (TriSAM)
Connectivity matrices
FlyWire v783					139,255 neurons	zenodo codex	synapse table and edge list of the whole adult fly brain; annotations at github
Witvliet connectomes					8 connectomes	zenodo nemanode	c. elegans developmental connectivity matrices (Witvliet et al. 2021)
L1EM adjacency					3,016 neurons 548,000 synapses	github	larval drosophila connectome matrices from supplementary of Winding et al. 2023
Proofreading and image restoration
ConnectomeBench2					716,485 decisions (>4.5 million images)	huggingface github	split/merge proofreading decisions across four open connectomes (mouse, human, zebrafish, fly); 2026
ConnectomeBench						huggingface github	segment typing, split and merge tasks from MICrONS and FlyWire for (multimodal) LLMs; NeurIPS 2025
EMDiffuse						zenodo github	paired noisy/clean and low/high-resolution mouse cortex EM for denoising, super-resolution and isotropic reconstruction
Multi-dataset collections
EMNeuron		multi (5–10 in-plane)			>3 of 22 billion voxels labeled	huggingface	16 sub-datasets (ZFinch, HBrain, FIB25, H01, Pinky, FAFB, Kasthuri, Basil, Harris, ...); multi-species/modality collection of SegNeuron (MICCAI 2024)

for pre-training at scale (unlabeled), see Pre-training Corpora

Pre-training Corpora

Large unlabeled (or weakly labeled) image collections for self-supervised and foundation-model training.

• CEM500K ^EMPIAR – ~500,000 curated 2D cellular EM patches (Conrad and Narayan 2021)
• CEM1.5M ^EMPIAR – 1,592,753 unlabeled 2D patches behind MitoNet (code)
• EM_pretrain_data ^TokenUnify – unlabeled neuron EM data used for TokenUnify pre-training

Data Portals and Tools

Repositories and portals

• BossDB ^{cloud repository for petascale EM and X-ray volumes (MICrONS, FANC, BANC, many published connectomes)}
• EMPIAR ^{EMBL-EBI archive for raw EM data, including many vEM entries}
• OpenOrganelle ^{Janelia FIB-SEM volumes of cells and tissues with organelle segmentations}
• neuPrint ^{connectome database and analysis service for the Janelia FlyEM datasets (hemibrain, MANC, optic lobe, male CNS)}
• FlyWire Codex ^{browser for the FlyWire whole-brain connectome and BANC}
• MICrONS Explorer ^{MICrONS data releases and tutorials}
• webKnossos ^{browser-based viewing and annotation; hosts many published datasets}
• Open Connectome Project ^{NeuroData's open EM connectomics archive}
• EBRAINS Knowledge Graph ^{European research infrastructure; human FIB-SEM synapse datasets}
• Zenodo and Hugging Face Datasets ^{paper-level releases of annotations and benchmarks}

Viewing, proofreading and analysis

• Neuroglancer ^{WebGL viewer for petascale volumes (precomputed, N5, Zarr)}
• CAVE ^{Connectome Annotation Versioning Engine behind FlyWire, MICrONS, FANC and BANC}
• CATMAID ^{collaborative skeleton tracing and connectivity annotation}
• VAST ^{volume annotation and segmentation tool}
• Paintera ^{dense label proofreading on N5/Zarr}
• CloudVolume and Igneous ^{read, write and downsample precomputed volumes}
• neuprint-python, CAVEclient, navis, natverse ^{programmatic access and neuron analysis}

Reviews

• Nature Methods editorial. Method of the Year 2025: Electron Microscopy-based Connectomics. 2025  ^collection
• Helmstaedter. Synaptic-resolution Connectomics: Towards Large Brains and Connectomic Screening. 2025
• Bock. Synaptic Connectomics: Status and Prospects. 2025
• Czymmek et al. Accelerating Data Sharing and Reuse in Volume Electron Microscopy. 2024
• Collinson et al. Volume EM: A Quiet Revolution Takes Shape. 2023
• Eisenstein. Seven Technologies to Watch in 2023. 2023  ^{volume EM is one of them}
• Jefferis et al. Scaling up Connectomics: The Road to a Whole Mouse Brain Connectome. 2023
• Peddie et al. Volume Electron Microscopy. 2022  ^{Nature Reviews Methods Primers}
• Kievits et al. How Innovations in Methodology Offer New Prospects for Volume Electron Microscopy. 2022
• Beyer et al. A Survey of Visualization and Analysis in High-Resolution Connectomics. 2022
• Galili et al. Connectomics and the Neural Basis of Behaviour. 2022
• Abbott et al. The Mind of a Mouse. 2020
• Motta et al. Big Data in Nanoscale Connectomics, and the Greed for Training Labels. 2019
• Kornfeld and Denk. Progress and Remaining Challenges in High-throughput Volume Electron Microscopy. 2018
• Schröter et al. Micro-connectomics: Probing the Organization of Neuronal Networks at the Cellular Scale. 2017
• Titze and Genoud. Volume Scanning Electron Microscopy for Imaging Biological Ultrastructure. 2016
• Lichtman et al. The Big Data Challenges of Connectomics. 2014
• Helmstaedter. Cellular-resolution Connectomics: Challenges of Dense Neural Circuit Reconstruction. 2013
• Briggman and Bock. Volume Electron Microscopy for Neuronal Circuit Reconstruction. 2012
• Lichtman and Denk. The Big and the Small: Challenges of Imaging the Brain's Circuits. 2011

Related Lists and Surveys

• connectomics-vis-survey.github.io
• braincircuits.io/resources
• tianyan.gitlab.io/braindata/connectomics-survey
• github.com/subeeshvasu/Awesome-Neuron-Segmentation-in-EM-Images
• github.com/Levishery/connectomic-paper-datasets
• volumeem.org: IntrovEM papers ^{reading list of the vEM community}
• The Adult Drosophila Connectome Ecosystem ^{how to access and cite the six adult fly connectomes (2026)}
• github.com/sjcabs/fly_connectome_data_tutorial ^{tutorial for the main fly connectomes}

Related Code

• Flood-Filling Networks ^{FFN segmentation (J0126, H01, mapzebrain, electrosensory lobe)}
• SOFIMA ^{scalable optical-flow alignment of EM sections}
• Local Shape Descriptors
• PyTorch Connectomics
• empanada / MitoNet ^{generalist mitochondria segmentation in napari}
• SynapseNet ^{vesicle and synapse segmentation}
• EMDiffuse ^{diffusion-based EM restoration and isotropic reconstruction}
• SegNeuron ^{generalist model + EMNeuron database}
• TokenUnify ^{autoregressive pre-training + Wafer (MEC) data}

Contribution

(including the private communication)

• Hao Zhai
• Liuyun Jiang
• Xinghui Zhao
• Yanchao Zhang
• Jinyue Guo

Please contribute ^{Pull Request & Issue} if you think a new dataset or a relevant paper is missing.

Let's enjoy the beauty of EM data and the awesome micro-connectomes!