SynQuant

SynQuant implements an unsupervised, probability-principled synapse detection algorithm to automate quantification of synapses from fluorescence microscopy images and control the false discovery rate.

Key Features:

• Probability-principled detection: Employs an algorithm derived from the theory of order statistics that controls the false discovery rate while enhancing detection power.
• Unsupervised operation: Performs synapse detection without prior training data or manual intervention.
• 2D and 3D compatibility: Processes both two-dimensional and three-dimensional imaging data.
• Multi-channel staining support: Handles multiple fluorescence staining channels common in synaptic imaging experiments.
• Robustness to imaging variability: Improves detection accuracy under heterogeneous brightness and low signal-to-noise ratio conditions.

Scientific Applications:

• Antibody specificity mitigation: Reduces misidentification caused by imperfect antibody specificity in synapse labeling.
• Heterogeneous brightness handling: Accounts for variations in antibody concentration and intrinsic differences among synaptic puncta.
• Low signal-to-noise imaging: Enhances synapse detection in images with low signal-to-noise ratios.
• Diverse imaging data: Applicable to data types including 3D in-vivo imaging, array tomography, and neuron-astrocyte coculture.

Methodology:

SynQuant integrates a novel synapse detection algorithm derived from the theory of order statistics and was validated on synthetic and real datasets with ground truth annotations or manual labels; it was shown to outperform specialized unsupervised synapse detection tools and generic spot detection methods across data types including 3D in-vivo imaging, array tomography, and neuron-astrocyte coculture.