Automatic detection of neurons in large cortical slices

Area of Science:

• Neuroscience
• Computational Biology
• Histology

Background:

• Analyzing neuron distribution in the cerebral cortex is crucial for understanding aging, disease, and neural coding.
• Manual neuron marking is time-consuming and a significant bottleneck in neuroscientific research.
• Existing methods lack efficiency for large-scale, high-resolution cortical slice analysis.

Purpose of the Study:

• To develop an automated system for analyzing neuron distribution in high-resolution cortical slices.
• To overcome the limitations of manual neuron identification methods.
• To enable efficient and accurate assessment of cortical structure and neuron populations.

Main Methods:

• A novel automated system that stitches tile images of cortical slices.
• Implementation of a multilayer shape analysis algorithm for reliable neuron identification.
• Processing of Nissl-stained human cortical slices (BA4 area) at high resolution (0.264 µm/pixel).

Main Results:

• The automated system successfully stitches tile images and identifies neurons.
• The neuron identification algorithm achieves an average accuracy of 87±6% with a 14±9% false positive rate.
• The method processes large, high-resolution cortical slices in a relatively short time.

Conclusions:

• This is the first automated algorithm for analyzing large, high-resolution cortical slices.
• The developed method significantly reduces the time and effort required for neuron distribution analysis.
• The approach facilitates advanced studies in cortical morphology, aging, and pathology.