Chunkflow: hybrid cloud processing of large 3D images by convolutional nets

Wu, Jingpeng; Silversmith, William; Lee, Ki-Suk; Seung, H. Sebastian

doi:10.1038/s41592-021-01088-5

Cited by 29 publications

(36 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A convolutional network was also trained to perform a semantic segmentation of the image for neurite classifications, including (1) soma+nucleus, (2) axon, (3) dendrite, (4) glia, and (5) blood vessel. Following the methods described in (Wu et al 2021), both networks were applied to the entire dataset at 8 × 8 × 40 nm 3 in overlapping chunks to produce a consistent prediction of the affinity and neurite classification maps. The segmentation output mask was applied to the predictions.…”

Section: Segmentationmentioning

confidence: 99%

“…A convolutional network was trained to predict whether a given voxel participated in a synaptic cleft. Inference on the entire dataset was processed using the methods described in (Wu et al 2021) using 8 × 8 × 40 nm 3 images. These synaptic cleft predictions were segmented using connected components, and components smaller than 40 voxels were removed.…”

Section: Synapse Detection and Assignmentmentioning

confidence: 99%

See 1 more Smart Citation

Functional connectomics spanning multiple areas of mouse visual cortex

Bodor¹,

Halageri²,

Sterling³

et al. 2021

Preprint

Self Cite

126

161

View full text Add to dashboard Cite

The value of an integrated approach for understanding the neocortex by combining functional characterization of single neuron activity with the underlying circuit architecture has been understood since the dawn of modern neuroscience. However, in practice, anatomical connectivity and physiology have been studied mostly separately. Following in the footsteps of previous studies that have combined physiology and anatomy in the same tissue, here we present a unique functional connectomics dataset that contains calcium imaging of an estimated 75,000 neurons from primary visual cortex (VISp) and three higher visual areas (VISrl, VISal and VISlm), that were recorded while a mouse viewed natural movies and parametric stimuli. The functional data were co-registered with electron microscopy (EM) data of the same volume which were automatically segmented, reconstructing more than 200,000 cells (neuronal and non-neuronal) and 524 million synapses. Subsequent proofreading of some neurons in this volume yielded reconstructions that include complete dendritic trees as well the local and inter-areal axonal projections. The largest proofread excitatory axon reached a length of 19 mm and formed 1,893 synapses, while the largest inhibitory axon formed 10,081 synapses. Here we release this dataset as an open access resource to the scientific community including a set of analysis tools that allows easy data access, both programmatically and through a web user interface.

show abstract

Section: Segmentationmentioning

confidence: 99%

Section: Synapse Detection and Assignmentmentioning

confidence: 99%

Functional connectomics spanning multiple areas of mouse visual cortex

Bodor¹,

Halageri²,

Sterling³

et al. 2021

Preprint

Self Cite

126

161

View full text Add to dashboard Cite

show abstract

“…Boundary detection, semantic labeling, and synaptic cleft detection model inference were implemented using Chunkflow (J. Wu et al 2021). This software package was developed previously for our terascale pipeline, and extended with only minor changes for the petascale dataset.…”

Section: Distributed Computationmentioning

confidence: 99%

“…Matching how we planned to process the larger volume, we merged connected components together which had centroid coordinates within 700 nm from one another and were assigned to the same synaptic partners in each training volume. This yielded cleft detection performance estimates of 98.2 precision and 98.2 recall for our small volume test set with 56 clefts.Following the methods described in (J Wu et al 2021),. we predicted synaptic clefts on the entire aligned dataset.…”

mentioning

confidence: 99%

Petascale neural circuit reconstruction: automated methods

Macrina

Lee

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

3D electron microscopy (EM) has been successful at mapping invertebrate nervous systems, but the approach has been limited to small chunks of mammalian brains. To scale up to larger volumes, we have built a computational pipeline for processing petascale image datasets acquired by serial section EM, a popular form of 3D EM. The pipeline employs convolutional nets to compute the nonsmooth transformations required to align images of serial sections containing numerous cracks and folds, detect neuronal boundaries, label voxels as axon, dendrite, soma, and other semantic categories, and detect synapses and assign them to presynaptic and postsynaptic segments. The output of neuronal boundary detection is segmented by mean affinity agglomeration with semantic and size constraints. Pipeline operations are implemented by leveraging distributed and cloud computing. Intermediate results of the pipeline are held in cloud storage, and can be effortlessly viewed as images, which aids debugging. We applied the pipeline to create an automated reconstruction of an EM image volume spanning four visual cortical areas of a mouse brain. Code for the pipeline is publicly available, as is the reconstructed volume.

show abstract

“…We used a combination of Neuroglancer (Maitin-Shepard, https://github.com/google/neuroglancer) and custom tools to annotate and store labeled spatial points 64 . In brief, we used Neuroglancer to simultaneously visualize the imagery and segmentation of the 3d EM data.…”

Section: Proofreading and Annotation Of Volumetric Imagery Datamentioning

confidence: 99%

Oligodendrocyte precursor cells prune axons in the mouse neocortex

Buchanan

Elabbady

Collman

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Neurons in the developing brain undergo extensive structural refinement as nascent circuits adopt their mature form. This transformation is facilitated by the engulfment and degradation of excess axonal branches and inappropriate synapses by surrounding glial cells, including microglia and astrocytes. However, the small size of phagocytic organelles and the complex, highly ramified morphology of glia has made it difficult to determine the contribution of these and other glial cell types to this process. Here, we used large scale, serial electron microscopy (ssEM) with computational volume segmentation to reconstruct the complete 3D morphologies of distinct glial types in the mouse visual cortex. Unexpectedly, we discovered that the fine processes of oligodendrocyte precursor cells (OPCs), a population of abundant, highly dynamic glial progenitors, frequently surrounded terminal axon branches and included numerous phagolysosomes (PLs) containing fragments of axons and presynaptic terminals. Single- nucleus RNA sequencing indicated that cortical OPCs express key phagocytic genes, as well as neuronal transcripts, consistent with active axonal engulfment. PLs were ten times more abundant in OPCs than in microglia in P36 mice, and declined with age and lineage progression, suggesting that OPCs contribute very substantially to refinement of neuronal circuits during later phases of cortical development.

show abstract

Chunkflow: hybrid cloud processing of large 3D images by convolutional nets

Cited by 29 publications

References 18 publications

Functional connectomics spanning multiple areas of mouse visual cortex

Functional connectomics spanning multiple areas of mouse visual cortex

Petascale neural circuit reconstruction: automated methods

Oligodendrocyte precursor cells prune axons in the mouse neocortex

Contact Info

Product

Resources

About