Research at Janelia using advanced imaging techniques at the light or ultrastructural level produces large volumes of complex image data that often require manual annotation, curation, and downstream analysis. Such annotations are also required to generate high-quality ground truth data for machine learning, with the goal of automating image processing and analysis at scale. Many of these efforts intersect with AI initiatives at Janelia/AI@HHMI.
The Bioimage Analysis team within Project Technical Resources (PTR) supports Janelia laboratories and project teams by providing structured, high-quality annotations across organisms, imaging modalities, and experimental contexts. We work closely with researchers to define annotation goals, establish consistent standards, and deliver datasets that enable quantitative analysis, machine-learning model development, and/or large-scale automation.
Fluorescent microscopy images
-
3D annotations of cells and subcellular structures (cell bodies, neurites, nuclei and other organelles) in fluorescent images generated by standard or expansion microscopy. The latter includes efforts to apply light-microscopy-based connectomics (LICONN) to fly, zebrafish, and mouse, harnessing information from both protein and RNA labeling at nanoscale resolution.
-
Time-resolved (4D) annotations capturing cell positions and dynamics over developmental time, enabling analysis of spatial organization and lineage relationships. Examples are the tracking of nuclei in C.elegans (Shroff lab) and following morphological changes in the developing mouse salivary gland (Wang lab).
-
Reconstruction of neurons in complex nervous systems, for example in mouse brain using methods established by MouseLight to trace complete axonal arbors or in the fly central nervous system using VVD viewer to establish correspondence between light-level imaging and electron microscopy datasets.
Ultrastructural image volumes
-
Voxel-level annotation of cellular structures in electron microscopy (FIB-SEM) datasets using Amira software, applying expertise generated by the Cellmap project team. Our current focus is on cell segmentations in various developmental stages in C. elegans in the context of the FuncEWOrm team project.
Animal body and organ models
PTR is involved in creating models of organs or organisms that are currently still poorly described. The Bioimage Analysis team collaborates with Igor Siwanowicz in PTR to model segmented structures in Blender. These projects emphasize accurate anatomy, spatial relationships, and organ-level structure. Examples include:
-
Organs and anatomical structures in microCT or expansion microscopy image volumes of Danionella (Satou lab).
-
Cellular composition of aphid salivary glands (Stern lab).
Animal behavior analysis
-
Animal part and pose tracking, including use of APT software (Branson lab) and other tracking frameworks.
-
Task- and lab-specific annotation workflows developed in collaboration with individual laboratories using custom tools for mouse, rat, and fly behavior (e.g., Tervo lab, Johnson lab).
Spike sorting curation
In addition to image and video annotations, PTR team members are trained in the curation of electrophysiology data recorded from multi-site electrodes (e.g., Neuropixels).