Braintown

Braintown automates identification and quantification of dendritic spines in live fluorescent tissue images using machine learning to assess structural synaptic plasticity.

Key Features:

• Automatic identification: Automatically detects dendritic spines within live fluorescent tissue images for high-throughput structural analysis.
• Machine learning-based image preprocessing: Applies custom thresholding and binarization functions to clean fluorescent images and enhance spine detectability.
• Neural network classification: Trains a neural network using features derived from the relative shape of the spine perimeter and its dendritic backbone to distinguish spines from other structures.
• High accuracy: Reports over 90% accuracy in detecting dendritic spines.

Scientific Applications:

• High-throughput screening: Screens numerous molecular targets for their effects on dendritic spine structural plasticity.
• Spine stimulation and monitoring: Supports stimulation and monitoring of hundreds of dendritic spines under varied experimental conditions in imaging studies.

Methodology:

Fluorescent images are prepared using custom thresholding and binarization functions; features based on the relative shape of the spine perimeter and its dendritic backbone are extracted; a neural network is trained on these features to identify dendritic spines and reduce manual parameter adjustments.