DeepCINAC

DeepCINAC infers neuronal activity from calcium imaging data using deep learning to resolve fluorescence transients and neuronal dynamics in densely packed, highly active regions such as the CA1 pyramidal layer of the hippocampus during early postnatal development.

Key Features:

• Deep Learning Architecture: Employs a convolutional neural network integrated with an attention mechanism and a bidirectional long-short term memory (LSTM) network to capture complex temporal patterns.
• Fluorescence Transient Detection: Detects fluorescence transients indicative of single-cell neuronal activity from calcium imaging data in densely packed regions.
• Performance Benchmarking: Achieves human-level performance and surpasses CaImAn in F1 score when evaluated against benchmark measurements.
• Automatic Identification of Neuronal Subtypes: Automatically identifies activity originating from specific neuronal subtypes such as GABAergic neurons.

Scientific Applications:

• Early postnatal hippocampal circuit analysis: Inferring single-cell activity in densely packed, synchronous networks such as the CA1 pyramidal layer during early postnatal development.
• Neurodevelopmental studies: Supporting investigations of neurodevelopmental processes through accurate single-cell activity inference.
• Synaptic plasticity research: Enabling analysis of activity patterns relevant to synaptic plasticity.
• Functional network organization: Characterizing the functional organization of neural networks from calcium imaging-derived activity.

Methodology:

Utilizes a convolutional neural network with an attention mechanism and a bidirectional LSTM to analyze calcium imaging data, with performance assessed using F1 score comparisons to CaImAn and human benchmarks.