NRRS: a re-tracing strategy to refine neuron reconstruction

Area of Science:

• Neuroscience
• Computational Biology
• Image Analysis

Background:

• Accurate neural morphology reconstruction from whole-brain images is essential for neuroscience research.
• Automated algorithms are needed to address challenges posed by large-scale, high-dimensional neuroimaging data.
• Existing methods struggle with deviations in reconstructed branches and bifurcation points.

Purpose of the Study:

• To develop and validate a novel automated strategy for refining neuron morphology reconstructions.
• To address deviation errors in large-scale neuron reconstruction datasets.
• To improve the accuracy of neuron skeleton representation, radius estimation, and axonal bouton detection.

Main Methods:

• Proposed a Neuron Reconstruction Refinement Strategy (NRRS) involving partitioning reconstructions into fixed-size segments.
• Implemented a two-step re-tracing process to resolve deviation errors.
• Validated the method using a synthetic dataset and applied it to the SEU-ALLEN/BICCN dataset.

Main Results:

• NRRS demonstrated superior performance compared to existing solutions in handling deviation errors.
• Achieved significant improvements in neuron skeleton accuracy, radius estimation, and axonal bouton detection on the SEU-ALLEN/BICCN dataset.
• The method successfully refined 1741 complete neuron reconstructions.

Conclusions:

• NRRS is a critical tool for enhancing the accuracy of neuron morphology reconstruction.
• The proposed strategy effectively refines reconstructions, contributing to more reliable large-scale neuroimaging analysis.
• Automated refinement is crucial for overcoming limitations in current neuron reconstruction techniques.