Open-Source Software for Real-time Calcium Imaging and Synchronized Neuron Firing Detection.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Bioengineering

Background:

• Real-time analysis of neural activity is crucial for understanding brain function.
• Existing calcium imaging analysis methods often lack real-time capabilities and struggle with accurate neuron identification.
• Closed-loop experiments are essential for establishing causal links between neural activity and behavior.

Purpose of the Study:

• To develop a real-time calcium imaging software (Carignan) for automatic detection of neural activity patterns.
• To enable closed-loop experiments by synchronizing neural activity detection with external device activation.
• To improve the accuracy of neuron detection using deep learning methods.

Main Methods:

• Development of Carignan software for real-time calcium imaging analysis.
• Implementation of a deep learning classification module using convolutional and recurrent neural networks (CNNs and RNNs) to distinguish true neurons from false positives.
• Integration with one-photon (1p) miniscopes for calcium imaging in freely-moving animals.
• Synchronization of neural activity detection with external device activation for optogenetics.

Main Results:

• Carignan successfully detects neural activity patterns in real-time.
• The deep learning module achieved superior performance in distinguishing neurons from false positives, evidenced by a high F1 score compared to existing methods.
• The software enables seamless integration with optogenetics for closed-loop experiments.

Conclusions:

• Carignan provides a robust platform for real-time neural activity analysis and closed-loop experiments.
• The software enhances the investigation of neuronal functions by enabling precise manipulation of neural circuits based on observed activity.
• Its application with 1p miniscopes in freely-moving animals broadens its utility across diverse research paradigms.