NeuroBlocks--Visual Tracking of Segmentation and Proofreading for Large Connectomics Projects

Area of Science:

• Neuroscience
• Computational Biology
• Data Visualization

Background:

• Connectomics research requires segmenting massive electron microscopy datasets to map neural connections.
• Manual and semi-automatic segmentation of these large volumes is complex, time-consuming, and involves multiple users and tools.
• Existing workflows lack integrated systems for managing, tracking, and auditing segmentation progress.

Purpose of the Study:

• To introduce NeuroBlocks, a novel visualization system for managing and tracking large-scale volumetric segmentation data in connectomics.
• To provide a multi-user, web-based platform that integrates diverse segmentation and analysis tools.
• To enhance provenance, accountability, and auditing for neuroscience data segmentation.

Main Methods:

• Designed NeuroBlocks based on an analysis of domain-specific tasks and challenges in volumetric data segmentation.
• Developed a multi-user, web-based application integrating manual/semi-automatic segmentation, proofreading, visualization, and analysis tools.
• Implemented features for tracking state, progress, and evolution of segmentation data, ensuring management and auditing capabilities.

Main Results:

• NeuroBlocks is the first system to integrate a heterogeneous set of tools for large-scale neuroscience segmentation.
• The system provides crucial support for management, provenance, accountability, and auditing of volumetric segmentation data.
• Demonstrated utility through two case studies focusing on different user roles in a real-world connectomics project.

Conclusions:

• NeuroBlocks addresses the challenges of managing and tracking progress in large-scale connectomics segmentation projects.
• The integrated, web-based system facilitates collaboration and oversight in neuroscience research.
• This visualization system improves the efficiency and reliability of reconstructing neural wiring diagrams.