Related Experiment Video
Updated: Aug 5, 2026
11:36
Analyzing Synaptic Modulation of Drosophila melanogaster Photoreceptors after Exposure to Prolonged Light
Published on: February 10, 2017
SynAPSeg: A novel dataset and image analysis framework for deep learning-based synapse detection and quantification
Pascal Schamber1, Sahana Darbhamulla1, Molly Boyer1
1Department of Neuroscience, Tufts University, Boston, Massachusetts, United States of America.
Plos Computational Biology
|July 29, 2026
Summary
SynAPSeg, a new deep learning tool, accurately quantifies synaptic puncta in dense neural tissue. This framework maps millions of synapses, revealing age-related changes in inhibitory neurons and advancing neuroscience research.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Biophysics
Background:
- Quantifying synaptic organization at circuit scales is a major challenge in neuroscience.
- Current deep learning tools often fail to segment synaptic puncta in dense tissue.
- There is a need for robust, scalable analysis methods for synaptic architecture.
Purpose of the Study:
- Introduce SynAPSeg, an open-source deep learning framework for synaptic puncta segmentation and quantification.
- Develop and validate deep learning models using a novel, large-scale instance segmentation dataset for synaptic puncta.
- Enable comprehensive analysis of synaptic architecture in both healthy and diseased brain states.
Main Methods:
- Developed SynAPSeg, an open-source framework integrating deep learning models for synaptic puncta analysis.
- Created and utilized the first large-scale, publicly available instance segmentation dataset for synaptic puncta.
- Trained deep learning models achieving expert-level performance on benchmark datasets.
- Integrated models into an interactive interface supporting multi-dimensional data and automated pipelines.
Main Results:
- SynAPSeg models achieved expert-level performance in segmenting synaptic puncta.
- Mapped nearly 4 million excitatory postsynaptic PSD95 puncta in dorsal hippocampus inhibitory interneurons.
- Identified regional differences in synapse properties within the hippocampus.
- Revealed a reduction in PSD95 density in aged CA1 parvalbumin (PV)-positive inhibitory neurons.
Conclusions:
- SynAPSeg offers a scalable solution for comprehensive synaptic architecture studies.
- The framework facilitates the discovery of age-associated synaptic alterations in inhibitory neurons.
- Findings suggest reduced glutamatergic recruitment of PV neurons in aging may contribute to cognitive decline.
- SynAPSeg advances the study of synaptic organization in neuroscience research.
