SyConn2: dense synaptic connectivity inference for volume electron microscopy

Schubert, Philipp J; Dorkenwald, Sven; Klimesch, Jonathan; Svara, Fabian; Mancu, Andrei; Ahmad, Hashir; Fee, Michale S.; Jain, Viren; Kornfeld, Joergen

doi:10.1038/s41592-022-01624-x

Cited by 14 publications

(14 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previous machine learning methods for neuropil annotation have primarily used features that were hand-designed or derived from supervised learning, including random forests trained on hand-designed features 9,21 , 2d convolutional networks trained on projections of neuropil ("Cellular Morphology Networks") 22,23 , point cloud networks trained on representations of cell membranes 24 , or 3d convolutional networks trained directly on voxels 25 . Schubert et al 22,24 trained cell representations using a triplet loss, but it was not reported whether these representations are suitable for downstream analyses. Previous results on cell type classification of neurite fragments required larger spatial context, precomputed organelle masks, and manually engineered features [21][22][23][24] , or used a single local view to achieve modest classification accuracy on a limited set of classes 22 .…”

Section: Introductionmentioning

confidence: 99%

“…Schubert et al 22,24 trained cell representations using a triplet loss, but it was not reported whether these representations are suitable for downstream analyses. Previous results on cell type classification of neurite fragments required larger spatial context, precomputed organelle masks, and manually engineered features [21][22][23][24] , or used a single local view to achieve modest classification accuracy on a limited set of classes 22 .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multi-Layered Maps of Neuropil with Segmentation-Guided Contrastive Learning

Dorkenwald

Berger³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Maps of the nervous system that identify individual cells along with their type, subcellular components, and connectivity have the potential to reveal fundamental organizational principles of neural circuits. Volumetric nanometer-resolution imaging of brain tissue provides the raw data needed to build such maps, but inferring all the relevant cellular and subcellular annotation layers is challenging. Here, we present Segmentation-Guided Contrastive Learning of Representations ("SegCLR"), a self-supervised machine learning technique that produces highly informative representations of cells directly from 3d electron microscope imagery and segmentations. When applied to volumes of human and mouse cerebral cortex, SegCLR enabled the classification of cellular subcompartments (axon, dendrite, soma, astrocytic process) with 4,000-fold less labeled data compared to fully supervised approaches. Surprisingly, SegCLR also enabled inference of cell types (neuron vs. glia and subtypes of each) from fragments with lengths as small as 10 micrometers, a task that can be difficult for humans to perform and whose feasibility greatly enhances the utility of imaging portions of brains in which many neuron fragments terminate at a volume boundary. These predictions were further augmented via Gaussian process uncertainty estimation to enable analyses restricted to high confidence subsets of the data.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Multi-Layered Maps of Neuropil with Segmentation-Guided Contrastive Learning

Dorkenwald

Berger³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Next we automated the detection of synaptic features. For this task, we used the SyConn v.2 pipeline 41,42 . We first used a deep neural network to map synaptic clefts and vesicle clouds (Fig.…”

Section: A Digital Address Book For All Neurons and Their Connectionsmentioning

confidence: 99%

“…We used SyConn2 (ref. 42 ) to process the automatic synaptic cleft, synaptic vesicle cloud and neurite segmentations and generate a synaptic connectome. The respective processing parameters and a source code snapshot can be found under https://gitlab.mpcdf.mpg.de/pschuber/ SyConn/-/tree/chunk_mask.…”

Section: Mapping Of Automatic Synapse Detection To Neurite Segmentationmentioning

confidence: 99%

Automated synapse-level reconstruction of neural circuits in the larval zebrafish brain

Svara¹,

Förster

Kubo

et al. 2022

Nat Methods

Self Cite

View full text Add to dashboard Cite

Dense reconstruction of synaptic connectivity requires high-resolution electron microscopy images of entire brains and tools to efficiently trace neuronal wires across the volume. To generate such a resource, we sectioned and imaged a larval zebrafish brain by serial block-face electron microscopy at a voxel size of 14 × 14 × 25 nm3. We segmented the resulting dataset with the flood-filling network algorithm, automated the detection of chemical synapses and validated the results by comparisons to transmission electron microscopic images and light-microscopic reconstructions. Neurons and their connections are stored in the form of a queryable and expandable digital address book. We reconstructed a network of 208 neurons involved in visual motion processing, most of them located in the pretectum, which had been functionally characterized in the same specimen by two-photon calcium imaging. Moreover, we mapped all 407 presynaptic and postsynaptic partners of two superficial interneurons in the tectum. The resource developed here serves as a foundation for synaptic-resolution circuit analyses in the zebrafish nervous system.

show abstract

“…This contrasts with the crucial functional reliance of neurons and synapses on their specific molecular machineries, reflected in enormous molecular diversity both at cellular and synaptic levels. EM reconstructions do allow inferring connectivity via chemical synapses from structural features with high accuracy (14)(15)(16)(17)(18). However, they are incomplete in the sense that in pure EM data, these cannot be further differentiated molecularly and information related to other forms of signaling between cells, like the distribution of receptor molecules beyond classical synaptic transmission, remains hidden.…”

mentioning

confidence: 99%

Light-microscopy based dense connectomic reconstruction of mammalian brain tissue

Tavakoli,

Lyudchik,

Januszewski

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

The information-processing capability of the brain’s cellular network depends on the physical wiring pattern between neurons and their molecular and functional characteristics. Mapping neurons and resolving their individual synaptic connections can be achieved by volumetric imaging at nanoscale resolution with dense cellular labeling. Light microscopy is uniquely positioned to visualize specific molecules but dense, synapse-level circuit reconstruction by light microscopy has been out of reach due to limitations in resolution, contrast, and volumetric imaging capability. Here we developed light-microscopy based connectomics (LICONN). We integrated specifically engineered hydrogel embedding and expansion with comprehensive deep-learning based segmentation and analysis of connectivity, thus directly incorporating molecular information in synapse-level brain tissue reconstructions. LICONN will allow synapse-level brain tissue phenotyping in biological experiments in a readily adoptable manner.One-Sentence SummaryHydrogel expansion enables molecularly informed reconstruction of brain tissue at synaptic resolution with light microscopy.

show abstract

SyConn2: dense synaptic connectivity inference for volume electron microscopy

Cited by 14 publications

References 32 publications

Multi-Layered Maps of Neuropil with Segmentation-Guided Contrastive Learning

Multi-Layered Maps of Neuropil with Segmentation-Guided Contrastive Learning

Automated synapse-level reconstruction of neural circuits in the larval zebrafish brain

Light-microscopy based dense connectomic reconstruction of mammalian brain tissue

Contact Info

Product

Resources

About