Training FlyWire AI Brain

Explain how electron-microscopy images, automated segmentation, human proofreading, and a connectome relate within FlyWire.

Configure the 2D imagery, segmentation overlays, coordinates, and 3D visibility for a no-edit inspection.

Navigate to a specified location and trace the same neurite through adjacent image sections.

Use image continuity and neuronal morphology to distinguish a likely continuation from a neighboring neurite.

Classify a candidate reconstruction problem as a false split, false merge, missing branch, or uncertain case.

Produce a concise candidate-problem note that records its location, visual evidence, uncertainty, and proposed next step.

Explain how segments, supervoxels, and root identifiers affect the scope and consequences of a FlyWire edit.

Verify a proposed edit by checking its evidence across adjacent sections and in the 3D reconstruction.

Add a disconnected segment to repair a well-supported false split.

Remove an incorrectly attached segment to repair a well-supported false merge.

Inspect the resulting morphology to detect unintended consequences of an edit.

Reverse or recover from an incorrect edit using the workflow available in the FlyWire interface.

Produce an audit log for a bounded proofreading case set that records edit, review, or unchanged decisions and their supporting evidence.

Configure an authenticated Python notebook and record the FlyWire release, materialization version, and coordinate system used for subsequent queries.

Resolve a neuron identifier against a specified materialized data version.

Retrieve annotations and basic metadata for a selected neuron.

Query the incoming and outgoing synapses of a selected neuron.

Use pandas to summarize synapse counts and rank the neuron's major connectivity partners by direction.

Assemble a reproducible single-neuron profile notebook containing a partner-distribution visualization, query parameters, and provenance information.

Diagram the FlyWire processing pipeline, distinguishing model training, segmentation inference, agglomeration, human proofreading, and connectome materialization.

Diagnose how image ambiguity can produce false-split and false-merge errors in example electron-microscopy regions.

Prepare a small pre-labeled electron-microscopy dataset as training and held-out arrays for a provided segmentation pipeline.

Train a lightweight baseline segmenter on the labeled image set.

Calculate intersection-over-union on the held-out images, inspect representative errors, and explain why this experiment illustrates—but does not retrain—FlyWire's production segmentation system.

Formulate a narrowly scoped connectomics question answerable with FlyWire data.

Select a small neuron cohort and record its identifiers, materialization version, selection criteria, and annotation provenance.

Validate the cohort's annotations and reconstructed morphology, documenting exclusions or unresolved cases.

Build a directed cohort-connectivity table containing synapse counts and in-degree and out-degree summaries.

Create a visualization that communicates the cohort’s strongest connections.

Evaluate which findings could change because of reconstruction uncertainty, annotation ambiguity, cohort selection, or data-version changes.

Package the capstone notebook, provenance information, findings, and limitations as a reproducible project.