Module 02: From Pixels to Projects — Exploring Real Connectomics Datasets
Dive into real-world brain mapping datasets and learn how EM images become 3D digital reconstructions of the brain.
Core Topics
- 🖼️ Volume Imaging: Acquiring nanoscale slices via serial-section EM
- 🔗 Image Alignment: Stitching and registering slices to form coherent volumes
- 🧩 Segmentation: Using AI to label pixels and define cell boundaries
- 🕸️ Skeletonization: Creating simplified graphs to represent neuron paths
- 🧠 Dataset Exploration: FlyWire, MICrONS, H01, FAFB, Kasthuri, and more
Sample Interactive Tools
COMPASS Integration
- Knowledge: Understand how data is organized and labeled for analysis
- Skills: Navigate viewers, identify structures, interpret metadata
- Character: Patience, attention to detail, and openness to uncertainty
- Meta-Learning: Learn how tools and pipelines evolve across datasets
References & Resources
- Kasthuri et al., 2015, Cell – "Saturated reconstruction of neocortex"
- Zheng et al., 2018, Nature – FlyWire / FAFB EM volume of adult fly brain
- MICrONS dataset release, 2021 – V1 Layer 2/3 mouse volume
- Allen Institute / BossDB
- BossDB Cookbook: Five-Minute Jump Start
- BossDB Cookbook: Advanced Data Downloading
Assessment
- Match dataset types to pipeline stages (e.g., segmentation → labeled volume)
- Navigate a dataset and find one synapse, one neuron, and one glial cell
- Write a 3–4 sentence description of how segmentation is performed and why it matters