Task Guides
Detailed guides for each annotation task type in the connectomics pipeline.
Proofreading (Tracing) — our core task TASK-01
What is Proofreading?
Proofreading is verifying and correcting AI-generated neuron segmentations. This is the team's namesake task — what Princeton Tracers were originally hired to do. We review automated reconstructions and fix errors where the AI incorrectly joined (merged) or separated (split) neurons.
Split/Merge Resolution is an integral part of proofreading. When two neurons are incorrectly joined, we perform a split operation. When a single neuron is fragmented into multiple segments, we perform a merge to reconnect them.
Why is it Important?
Automated segmentation algorithms are powerful but imperfect. False mergers and false splits in 3D neuron reconstructions can lead to incorrect connectivity data. Human proofreaders catch errors that AI cannot, ensuring the resulting connectome data is accurate enough for scientific analysis.
Three-Round Triage System
- Quick Scan — Rapidly scroll through the neuron looking for obvious errors: large mergers, clearly detached fragments, and branches that don't belong. Flag anything suspicious.
- Systematic Review — Work through each branch methodically. Check that branches originate from the correct soma, verify that thin processes aren't artifacts, and examine crossing fibers carefully for false mergers.
- Verification — Confirm corrections are clean. Check that splits haven't orphaned legitimate branches. Verify merges haven't introduced new errors. Review flagged regions one final time.
Key Operations
- Split — Separate incorrectly merged neurons at the point of error
- Merge — Reconnect fragments of the same neuron
- Undo — Reverse an incorrect operation
- Flag — Mark ambiguous regions for expert review
Common Failure Modes
- Missing small branches during quick scan
- Accepting AI-generated mergers without verifying at branch points
- Over-splitting at ambiguous or low-contrast regions
- Ignoring crossing fibers where two neurons pass near each other
- Rushing Round 1 and missing errors that compound later
Tools Used
Related Documentation
Semantic Segmentation — painting & labeling structures TASK-02
What is Semantic Segmentation?
Semantic segmentation is the process of labeling tissue types and cellular structures in electron microscopy data. This encompasses two related activities: voxel painting (hand-painting specific structures in 3D) and classification labeling (identifying and naming structures like axons, dendrites, and somata).
We have a storied history of hand-painting neurons, mitochondria, vesicles, synapses, T-bars, errors, membranes, and more. The tools and software have changed rapidly over the years as technology capabilities evolve, but the core skill of identifying and labeling biological structures in EM imagery remains the same.
Structure Labels
Note that the distinction between painting and labeling can be confusing for newer tracers. Painting means coloring in the actual voxels of a structure. Labeling means assigning a classification name (like "axon" or "dendrite") to an already-segmented object.
Defect Annotation
Defect annotation is a specialized form of semantic segmentation focused on marking imaging artifacts, tissue damage, and data quality issues. Defects include section tears, folds, staining artifacts, missing slices, and other preparation problems. Accurate defect annotation prevents these regions from corrupting downstream analysis. See the Gallery for visual examples of common defects.
Procedure
- Load the assigned volume or task in the annotation tool
- Orient yourself — identify major landmarks and the region of interest
- Select the appropriate label/brush for the structure you're painting
- Paint or label structures systematically, working through Z-slices
- Cross-check your labels in orthogonal views (XY, XZ, YZ) for consistency
- Submit for review when complete
Common Failure Modes
- Confusing axons with dendrites in ambiguous cross-sections
- Missing myelin wrapping around processes
- Inconsistent labeling across Z-slices (label drift)
- Guessing in damaged or low-contrast regions instead of flagging
- Boundary leakage when painting — extending labels past structure edges
Tools Used
Related Documentation
Skeletonization — neuron morphology tracing TASK-03
What is Skeletonization?
Skeletonization creates skeletal (centerline) representations of neuron morphology from 3D segmentations. These skeletons capture the branching structure and spatial extent of neurons in a compact format useful for analysis.
This is not a frequent manual task — skeletonization is usually automated using algorithms like TEASAR or Kimimaro. When manual work is needed, it typically involves reviewing and correcting automated skeletons rather than tracing from scratch.
When Manual Correction is Needed
- Automated skeleton missed thin processes or terminal branches
- Incorrect branch point topology (wrong connections)
- Centerline drifted outside the segmentation in curved regions
- Skeleton extends beyond the actual segmentation boundary
- Node density is too sparse to capture morphological features
Tools Used
Output formats: SWC, JSON, Precomputed meshes