DeNeRD: high-throughput detection of neurons for brain-wide analysis with deep learning

Area of Science:

• Neuroscience
• Artificial Intelligence
• Bioinformatics

Background:

• Mapping mammalian brain structure at cellular resolution is complex.
• Existing methods struggle with high-complexity data and whole-brain analysis.
• A meso-scopic approach is needed for cellular-level whole-brain mapping.

Purpose of the Study:

• Introduce a novel, automated AI-based method for whole-brain image processing.
• Develop a tool to Detect Neurons in different brain Regions during Development (DeNeRD).
• Address limitations of current cell detection methods in complex brain data.

Main Methods:

• Utilized a fully automated Artificial Intelligence (AI) deep neural network.
• Applied the method to whole-brain image processing.
• Tested neuron detection across diverse brain regions and developmental stages.

Main Results:

• Demonstrated high performance in detecting neurons.
• Successfully identified neurons labeled with various genetic markers.
• Validated the method across different imaging planes and modalities.

Conclusions:

• The AI-based DeNeRD method offers an effective solution for cellular-resolution whole-brain mapping.
• This approach overcomes significant constraints of existing cell detection techniques.
• DeNeRD shows high accuracy and versatility for neuroscience research.