SynapseNet: Deep Learning for Automatic Synapse Reconstruction
Posted on 24 February 2025
Electron microscopy provides unmatched detail for studying synaptic morphology, but manual segmentation is a bottleneck. SynapseNet is a deep-learning tool that automates not only synapse reconstruction but also efficient, scalable analysis. It is trained on a large annotated room temperature EM dataset and can segment structures such as synaptic vesicles (SV), mitochondria, active zones (AZ), and compartments. With domain adaptation functionality, it also reliably segments SVs across a wide range of imaging techniques and data domains.
For more details on SynapseNet check out the software at https://github.com/computational-cell-analytics/synapse-net and the preprint at https://doi.org/10.1101/2024.12.02.626387.
Contact email: constantin.pape@informatik.uni-goettingen.de
Resource Created By: Sarah Muth, Frederieke Moschref, Luca Freckmann, Sophia Mutschall, Ines Hojas-Garcia-Plaza, Julius N. Bahr, Arsen Petrovic, Thanh Thao Do, Valentin Schwarze, Anwai Archit, Kirsten Weyand, Susann Michanski, Lydia Maus, Cordelia Imig, Nils Brose, Carolin Wichmann, Ruben Fernandez-Busnadiego, Tobias Moser, Silvio O. Rizzoli, Benjamin H. Cooper, Constantin Pape
Type of tool: Image analysis software
Is this tool open source?: Yes
You may also be interested in
Advanced Cell Classifier
Machine learning method for image data analysis
Horvath Lab
SIMcheck
ImageJ tools for assessing SIM data quality
Micron Oxford
OpenSPIM
A platform to build, adapt and enhance SPIM technology