NeuroNex: 3DEM

Type: Software,

Keywords: 3DEM, BRAIN Initiative, Hippocampal Neuropil, NeuroNex, Neuroscience, Reconstruction tool

Neuropil tools for enhanced resolution 3-dimensional electron microscopy

Neuropil tools to edit 3DEM objects and SWiFT-IR for automated alignment. As a part of our NeuroNex grant, our collaborators at the SALK (Sejnowski lab) have created a tool (Neuropil Tools) that generates realistic meshes for objects derived from 3DEM data, and allows us to measure and calculate accurate surface areas and volumes of any object (eg dendritic spines, axonal boutons, and synapse surface areas). Our lab has provided the data and user feedback to make this tool accessible for novice users. The tool will be hosted on our website,, where it will be open source for the community to use.

* Web-based research platform.
* Focused on developing and disseminating new technologies for enhanced resolution 3-dimensional electron microscopy.
* 3DEM has become recognized as an important tool to map and understand synaptic circuitry in the brain.
* Enhanced resolution for 3DEM analysis of synapses across brain regions and taxa.
* Blender addon created to segment, annotate, and analyze the spines in the 3D reconstructions.
* Provide standards for identifying ultrastructural objects in 3DEM, realistic reconstructions for modeling biophysical properties of synaptic transmission, and a test bed for enhancing reconstruction tools.
* Obtaining detailed representations of the structural organization of dendrites, axons, glia, and their subcellular components is fundamental to understanding cellular mechanisms of behavior, learning, and memory.
* Results from 3DEM are beginning to provide crucial insights into the anatomical substrates of information processing and behavioral output in normal human brains and in pathological conditions such as Alzheimer’s disease.
* All three annotated series and associated Reconstruct files are publicly available for viewing and download at the Open Connectome Project.
* Web services can be accessed programmatically within user-facing analysis tools, such as Python, Matlab, and R, which allows scientists to access data and metadata interactively through their preferred analytics platform.
* Densely reconstructed volumes of hippocampal neuropil: provide accurate identification of axonal, dendritic, and glial processes as well as example subcellular dendritic structures to train students and researchers who are new to ultrastructural analysis.
* Original images are provided at high resolution without superimposed alignment scaling, so that alternative approaches can be tested.
* Provide a crucial basis from which to generate and test the validity of computer segmentation algorithms, to reduce operator involvement, and to provide accurate densely reconstructed volumes more quickly.
* Such automated discovery tools should greatly facilitate future experiments, which will elucidate how specific structures become altered during normal processes (such as learning) or in disease states (such as Alzheimer’s or epilepsy) and then provide new and more specific targets for treatment.
* Re-engineered software (e.g. Reconstruct and neuropil tools) to better handle larger datasets, and to be accessible via web browser.
* A dissemination portal that will allow access to a 3DEM brain image data repository for neuroscientists, researchers, educators, and students. The portal is integrated with high performance computing resources to allow data access and analysis in a collaborative workspace.

* Evaluates the axon-spine coupling in a complete nanoconnectomic three-dimensional reconstruction from serial electron microscopy (3DEM) of hippocampal neuropil
* Three volumes of hippocampal CA1 neuropil from adult rat were imaged at X-Y resolution of ~2 nm on serial sections of ~50–60 nm thickness. These are the first densely reconstructed hippocampal volumes

* High-resolution images will serve to develop machine learning algorithms that can automatically identify structurally distinct objects, such as the spine apparatus, that are highly important for local synapse function but occur infrequently in the neuropil
* New tools will also help to identify locations of ambiguity on the images and help to improve the reliability of reconstructions and surface measurements and renderings

* Even with the VolRover improvements 3DEM likely underestimates the extracellular space (ECS) because accurate determination of membrane boundaries can be especially difficult where membranes are obliquely sectioned. ECS can be further reduced by the chemical fixation, which causes astroglial processes to swell relative to cryo-electron microscopy. The degree to which ECS was lost versus obscured by overlying oblique membranes is an open question whose answer will require future 3D tomographic reconstructions, which provide z resolutions on virtual image sections in the range of 2–3 nm

Harris et al. 2015, A resource from 3D electron microscopy of hippocampal neuropil for user training and tool development, Scientific Data 2:150046.


Tom Bartol (Postdoc)
Lyndsey kirk (Postdoc)
Kristen Harris (Professor)


University of Texas at Austin