Method for counting labeled neurons in mouse brain regions based on image representation and registration

Area of Science:

• Neuroscience
• Computer Vision
• Biomedical Imaging

Background:

• Accurate quantification of labeled neurons within specific brain regions is crucial for statistical analysis in brain image analysis.
• Challenges in mouse brain slice analysis include noise, distortion, and modal differences with standard atlases, hindering precise region correspondence and accurate neuron counting.

Purpose of the Study:

• To develop an automated method for accurate counting of labeled neurons in mouse brain regions.
• To address inaccuracies in neuron counts caused by imprecise brain region segmentation in deformed brain slices.

Main Methods:

• Utilized image representation and neural networks for registration between different modal mouse brain slices and a standard brain atlas.
• Implemented regional localization of brain slices based on the registration.
• Employed threshold segmentation for detecting and counting labeled neurons within identified brain regions.

Main Results:

• The proposed method effectively overcomes the large deviations in neuron counts resulting from inaccurate brain region division in highly deformed brain slices.
• Demonstrated automatic and accurate counting of labeled neurons in each brain region of the analyzed slices.

Conclusions:

• The developed framework based on image representation and registration enables precise segmentation and accurate neuron quantification in mouse brain slices.
• This automated approach significantly improves the reliability of neuron counting for downstream statistical analysis in neuroscience research.