07 Glia

Session outcomes (60 minutes)

• Distinguish major glial classes in EM-oriented workflows.
• Reduce glia-neuron boundary errors in proofreading.
• Prioritize glia-related corrections by downstream impact.

Pedagogical arc

• Concept refresh: glia as active circuit context.
• Visual discrimination: class-specific cues.
• Practice: ambiguity triage and escalation.
• Check: class call plus uncertainty and action plan.

Why this matters technically

• Glia errors can induce merge/split cascades.
• Boundary mistakes alter neuron-centric metrics.
• Myelin and glial context changes interpretation of nearby neurites.

Real data: an astrocytic process

• Pale cytoplasm, irregular space-filling outline that wraps its neighbours.
• Labelled 'astrocyte' by H01's subcompartment model.

Real data: cortical vasculature

• H01 contains ~230 mm of blood vessels across ~1 mm³.
• Vessels are a common site of merge errors — glia wrap them closely.

Oligodendrocyte-associated cue context

• Diagnostic feature: the darkest nucleus in the field — small, round, extremely
  electron-dense heterochromatin, identifiable at low magnification.
• Dense cytoplasm: abundant rough ER and ribosomes, prominent Golgi. Soma ~6–8 µm.
• Processes connect the soma to myelin sheaths; one cell myelinates on the order of
  20–60 axonal segments.
• The hard case is not the mature cell. It is the OPC (NG2 cell): paler nucleus, can
  resemble a small neuron or an astrocyte. "Sort of oligodendrocyte but the nucleus is not
  dark enough" means OPC is the leading hypothesis — flag it, do not force the call.

No extracted micrograph exists for this class; see Unit 07 §2 for the reference images.

Myelin-context interpretation

• Myelin is the fastest route to an oligodendrocyte identification: follow a sheath back to
  the soma that owns it.
• Instructor cue: ask how myelin context changes proofreading priority. A merge across a
  myelinated axon costs more than one across an unmyelinated process, because the axon it
  corrupts is long-range.

Operational glia triage protocol

1. Identify likely class from morphology/context.
2. Validate local boundary integrity.
3. Estimate downstream risk if left uncorrected.
4. Route to immediate correction or adjudication queue.

Metrics to track

• Glia-neuron boundary error rate.
• Class-specific agreement across reviewers.
• Escalation rate and resolution time.
• Downstream correction impact on network summaries.

Misconceptions to correct

• "Glia are background, neuron labels matter more."
• "Any dark process near myelin is neuronal."
• "Class call can be deferred indefinitely without impact."

Activity

Classify two ambiguous glia-neuron interfaces and submit:

• class hypothesis,
• boundary-confidence score,
• correction priority rank,
• escalation note if unresolved.

Rubric checkpoint

• Pass: class + boundary rationale + action path.
• Strong: explicit risk prioritization and uncertainty language.
• Flag: class label without boundary logic.

External paper figure integration

• Connectomics papers with glia-aware reconstruction examples.
• Myelination and ultrastructure review figures for context.
• Dataset-specific glia annotation benchmark figures when available.

External inserted figure (open license)

• Source URL: https://commons.wikimedia.org/wiki/Special:FilePath/Glial_Cell_Types.png
• License: CC BY 3.0 Unported.

References and attribution

• Internal visuals: Pat Rivlin glia training set.
• Suggested supporting review: Harris & Weinberg (2012) for synaptic neighborhood context.