Build scientific and technical fluency in nanoscale connectomics from motivation through analysis methods.
This track builds the foundational knowledge and technical skills needed to work with nanoscale connectomics data. Starting from scientific question-framing and progressing through neuroanatomy, EM imaging, segmentation, and circuit analysis, it provides the conceptual toolkit for working with Mouse Connectome Project and other CONNECTS datasets. It maps directly onto the Technical Training Course units and is the recommended starting point for most learners in the program.
Fadel alignment: Knowledge, Skills
Who this is for. Readers like Julian, a first-generation undergraduate with no lab experience yet, and Maya, a graduate student crossing into connectomics from another field. Start here if you cannot yet read an EM image or say why a segmentation is wrong.
Time: 60-70 hours of focused work, typically over 12-16 weeks. Starting point: None. Introductory biology and comfort with basic arithmetic are enough to start.
Completion here is a capability, not a set of pages visited. You are through this track when:
The predictable ways learners lose time on this track:
The topic is the same in all three. What changes is the sequencing, who supplies feedback, and what counts as done — see modes for the full description of each.
Read the nine technical units in order, do every lab, and treat units 05-07 as the center of gravity rather than a descriptive interlude. Roughly 31 hours across the nine units and their labs; the module library on top of that is what takes this track to 60-70.
Worth knowing: The perceptual units are the part that most needs a second pair of eyes. Find one partner before you reach Unit 05.
Every module has a session kit. A common shape is a five-session block: framing, imaging, reading the images (two sessions), then reconstruction infrastructure.
Worth knowing: The two image-reading sessions carry the load. Do not compress them to make room for infrastructure.
Would sit downstream of this track: the units become the reference a trainee consults while working, and the calibration gate replaces the self-scored drills.
Turn broad interest in brain mapping into concrete, testable connectomics questions with explicit scope and measurable outcomes.
Hands-on Python and Jupyter skills for reproducible connectomics data exploration, from environment setup through documented analysis workflows.
Neuroanatomical fluency for interpreting EM structures across cortical layers and brain regions, with attention to uncertainty and misclassification risks.
How EM produces the raw data of connectomics: acquisition principles, common artifacts, and image quality screening for segmentation readiness.
Core segmentation error taxonomy—merges, splits, boundary errors—and a practical correction workflow with documented quality impact.
Proofreading strategies that prioritize scientifically high-impact corrections and maintain reproducible, documented QC standards.
Designing testable connectomics hypotheses with measurable structural outcomes, appropriate null models, and explicit uncertainty limits.
Extracting and interpreting skeleton representations and morphology descriptors from segmented neurons for cell-type reasoning.
Representing connectomes as graphs, computing core network metrics, and interpreting results with biological and statistical caution.
Interpreting synaptic organization and local circuit motifs from connectomics data, differentiating robust patterns from reconstruction artifacts.
Canonical connectomics sequence from imaging to NeuroAI.
Per-unit build plans for instructors: slide sequence, timing, figure placement, and speaker notes, with links to the rendered Marp decks.
Required and optional readings aligned to the technical sequence.
Shared vocabulary for methods, quality, and analysis.
Data resources for training and research, including the MouseConnects dataset.
Filter concepts by immediate need to find the most relevant next resources.
Track: core-concepts-methods
User needs: starting a research question, avoiding overclaiming
Translate broad brain questions into testable structural hypotheses with clear evidence boundaries.
How to learn it: Start with one biological question, define measurable structural outputs, then state explicit non-claims.
Teaching set:
Track: core-concepts-methods
User needs: matching method to question, planning compute and storage
Choose imaging and analysis scale that can resolve required features at manageable cost.
How to learn it: Match your hypothesis to the smallest sufficient resolution and volume, then budget compute before data acquisition.
Teaching set:
Track: core-concepts-methods
User needs: improving data quality, debugging acquisition issues
Identify acquisition artifacts and define practical QA gates before reconstruction.
How to learn it: Use a shared artifact taxonomy and pass/fail thresholds so acquisition issues are caught before reconstruction.
Teaching set:
Track: core-concepts-methods
User needs: building robust pipelines, reproducible processing
Design scalable ingest-to-serving systems with lineage, release, and rollback discipline.
How to learn it: Treat reconstruction as production infrastructure: lineage, observability, and rollback are core scientific requirements.
Teaching set:
Track: core-concepts-methods
User needs: reading EM confidently, improving annotation consistency
Use compartment, organelle, and synaptic cues to make reproducible interpretation decisions.
How to learn it: Combine multiple ultrastructural cues across slices and label uncertainty explicitly instead of forcing hard calls.
Teaching set:
Track: core-concepts-methods
User needs: reducing identity confusion, handling ambiguity
Apply multi-cue decision rules and confidence tags for process-type calls.
How to learn it: Use morphology plus context cues to classify neurites and escalate ambiguous cases consistently.
Teaching set:
Track: core-concepts-methods
User needs: reducing glia-neuron boundary errors, interpreting myelin context
Distinguish major glia classes and integrate glia decisions into high-value QC workflows.
How to learn it: Focus on glial ultrastructure signatures and boundary integrity to reduce high-impact proofreading errors.
Teaching set: