In silico neuronal morphology classification: A systematic review

Area of Science:

• Neuroscience
• Computational Biology
• Bioinformatics

Background:

• Understanding brain function relies on connectomics and neuronal diversity.
• Classifying neuronal morphology is crucial but computationally complex.
• A systematic review is needed to map current methods and future trends.

Purpose of the Study:

• To review state-of-the-art computational methods for neuronal morphology classification.
• To identify common data representations and evaluation metrics.
• To provide an updated perspective guiding future neuroscience research.

Main Methods:

• Systematic review of peer-reviewed studies from 2018-2024.
• Searched five databases, collecting 840 papers.
• Assessed 35 studies for quality, methodology, and reporting.

Main Results:

• Artificial Neural Networks are the dominant computational method (21 papers).
• Microscopy images are the most common data representation (30 papers).
• Accuracy is the predominant evaluation measure (29 papers).

Conclusions:

• Artificial Neural Networks and image-based features are key in current neuronal classification.
• Standardized evaluation metrics are crucial for progress.
• This review offers a roadmap for future research in neuronal morphology classification.